refactor(fusion_accounting): move AI module code into fusion_accounting_ai sub-module

git mv preserves history. fusion_accounting/ retains only __manifest__.py,
__init__.py, CLAUDE.md, and docs/ — the meta-module shell. All Python,
data, views, security, services, static, tests, wizards, report move to
fusion_accounting_ai/. Manifest data list updated; security.xml move to
_core deferred to Task 12.

Made-with: Cursor

fusion_accounting_ai/services/__init__.py

@@ -0,0 +1,5 @@
+from . import adapters
+from . import tools
+from . import prompts
+from . import agent
+from . import scoring

fusion_accounting_ai/services/adapters/__init__.py

@@ -0,0 +1,2 @@
+from . import claude
+from . import openai_adapter

fusion_accounting_ai/services/adapters/claude.py

@@ -0,0 +1,141 @@
+import json
+import logging
+
+from odoo import models, api, _
+from odoo.exceptions import UserError
+
+_logger = logging.getLogger(__name__)
+
+try:
+    import anthropic as anthropic_sdk
+except ImportError:
+    anthropic_sdk = None
+
+
+class FusionAccountingAdapterClaude(models.AbstractModel):
+    _name = 'fusion.accounting.adapter.claude'
+    _description = 'Claude AI Adapter'
+
+    def _get_client(self):
+        if anthropic_sdk is None:
+            raise UserError(_("The 'anthropic' Python package is not installed."))
+        try:
+            key = self.env['fusion.api.service'].get_api_key(
+                provider_type='anthropic',
+                consumer='fusion_accounting',
+                feature='chat_with_tools',
+            )
+        except Exception:
+            ICP = self.env['ir.config_parameter'].sudo()
+            key = ICP.get_param('fusion_accounting.anthropic_api_key', '')
+            if not key:
+                raise UserError(_("No Anthropic API key configured."))
+        return anthropic_sdk.Anthropic(api_key=key)
+
+    def _get_model_name(self):
+        ICP = self.env['ir.config_parameter'].sudo()
+        return ICP.get_param('fusion_accounting.claude_model', 'claude-sonnet-4-6')
+
+    def _format_tools(self, tools):
+        formatted = []
+        for tool in tools:
+            t = {
+                'name': tool['name'],
+                'description': tool['description'],
+                'input_schema': tool.get('parameters', {'type': 'object', 'properties': {}}),
+            }
+            formatted.append(t)
+        return formatted
+
+    def _supports_extended_thinking(self, model):
+        return '4-6' in model or '4-5' in model or '4-1' in model or '4-0' in model
+
+    def call_with_tools(self, system_prompt, messages, tools=None, model_override=None):
+        client = self._get_client()
+        model = model_override or self._get_model_name()
+
+        api_messages = []
+        for msg in messages:
+            if msg['role'] in ('user', 'assistant'):
+                api_messages.append(msg)
+
+        kwargs = {
+            'model': model,
+            'max_tokens': 16384,
+            'system': system_prompt,
+            'messages': api_messages,
+        }
+        if tools:
+            kwargs['tools'] = self._format_tools(tools)
+
+        if self._supports_extended_thinking(model) and tools:
+            kwargs['thinking'] = {
+                'type': 'enabled',
+                'budget_tokens': 8192,
+            }
+
+        try:
+            response = client.messages.create(**kwargs)
+        except anthropic_sdk.BadRequestError as e:
+            if 'thinking' in str(e).lower():
+                kwargs.pop('thinking', None)
+                response = client.messages.create(**kwargs)
+            else:
+                raise UserError(_("Claude API error: %s", str(e)))
+        except Exception as e:
+            _logger.error("Claude API error: %s", e)
+            raise UserError(_("Claude API error: %s", str(e)))
+
+        text_parts = []
+        tool_calls = []
+        thinking_text = []
+        for block in response.content:
+            if block.type == 'text':
+                text_parts.append(block.text)
+            elif block.type == 'tool_use':
+                tool_calls.append({
+                    'id': block.id,
+                    'name': block.name,
+                    'arguments': block.input,
+                })
+            elif block.type == 'thinking':
+                thinking_text.append(block.thinking)
+
+        if thinking_text:
+            _logger.debug("Claude thinking: %s", thinking_text[0][:500])
+
+        return {
+            'text': '\n'.join(text_parts),
+            'tool_calls': tool_calls if tool_calls else None,
+            'tokens_in': response.usage.input_tokens,
+            'tokens_out': response.usage.output_tokens,
+            'stop_reason': response.stop_reason,
+            'raw_content': response.content,
+        }
+
+    def append_tool_results(self, messages, ai_response, tool_results):
+        assistant_content = []
+        for block in ai_response.get('raw_content', []):
+            if hasattr(block, 'type'):
+                if block.type == 'text':
+                    assistant_content.append({'type': 'text', 'text': block.text})
+                elif block.type == 'tool_use':
+                    assistant_content.append({
+                        'type': 'tool_use',
+                        'id': block.id,
+                        'name': block.name,
+                        'input': block.input,
+                    })
+
+        messages.append({'role': 'assistant', 'content': assistant_content})
+
+        tool_result_content = []
+        for tr in tool_results:
+            tool_result_content.append({
+                'type': 'tool_result',
+                'tool_use_id': tr['tool_call_id'],
+                'content': tr['result'],
+            })
+        messages.append({'role': 'user', 'content': tool_result_content})
+
+        return messages

fusion_accounting_ai/services/adapters/openai_adapter.py

@@ -0,0 +1,137 @@
+import json
+import logging
+
+from odoo import models, api, _
+from odoo.exceptions import UserError
+
+_logger = logging.getLogger(__name__)
+
+try:
+    from openai import OpenAI
+except ImportError:
+    OpenAI = None
+
+
+class FusionAccountingAdapterOpenAI(models.AbstractModel):
+    _name = 'fusion.accounting.adapter.openai'
+    _description = 'OpenAI AI Adapter'
+
+    def _get_client(self):
+        if OpenAI is None:
+            raise UserError(_("The 'openai' Python package is not installed."))
+        try:
+            key = self.env['fusion.api.service'].get_api_key(
+                provider_type='openai',
+                consumer='fusion_accounting',
+                feature='chat_with_tools',
+            )
+        except Exception:
+            ICP = self.env['ir.config_parameter'].sudo()
+            key = ICP.get_param('fusion_accounting.openai_api_key', '')
+            if not key:
+                raise UserError(_("No OpenAI API key configured."))
+        return OpenAI(api_key=key)
+
+    def _get_model_name(self):
+        ICP = self.env['ir.config_parameter'].sudo()
+        return ICP.get_param('fusion_accounting.openai_model', 'gpt-5.4-mini')
+
+    def _format_tools(self, tools):
+        formatted = []
+        for tool in tools:
+            formatted.append({
+                'type': 'function',
+                'function': {
+                    'name': tool['name'],
+                    'description': tool['description'],
+                    'parameters': tool.get('parameters', {'type': 'object', 'properties': {}}),
+                },
+            })
+        return formatted
+
+    def _is_reasoning_model(self, model):
+        return model.startswith('o1') or model.startswith('o3') or model.startswith('o4')
+
+    def call_with_tools(self, system_prompt, messages, tools=None, model_override=None):
+        client = self._get_client()
+        model = model_override or self._get_model_name()
+        is_reasoning = self._is_reasoning_model(model)
+
+        if is_reasoning:
+            api_messages = [{'role': 'developer', 'content': system_prompt}]
+        else:
+            api_messages = [{'role': 'system', 'content': system_prompt}]
+        for msg in messages:
+            if msg['role'] in ('user', 'assistant', 'tool'):
+                api_messages.append(msg)
+
+        kwargs = {
+            'model': model,
+            'messages': api_messages,
+        }
+        if is_reasoning:
+            kwargs['max_completion_tokens'] = 16384
+            kwargs['reasoning_effort'] = 'medium'
+        else:
+            kwargs['max_tokens'] = 4096
+
+        if tools:
+            kwargs['tools'] = self._format_tools(tools)
+
+        try:
+            response = client.chat.completions.create(**kwargs)
+        except Exception as e:
+            _logger.error("OpenAI API error: %s", e)
+            raise UserError(_("OpenAI API error: %s", str(e)))
+
+        choice = response.choices[0]
+        message = choice.message
+
+        tool_calls = []
+        if message.tool_calls:
+            for tc in message.tool_calls:
+                try:
+                    args = json.loads(tc.function.arguments)
+                except (json.JSONDecodeError, TypeError):
+                    _logger.warning("Malformed tool args from OpenAI: %s", tc.function.arguments)
+                    args = {}
+                tool_calls.append({
+                    'id': tc.id,
+                    'name': tc.function.name,
+                    'arguments': args,
+                })
+
+        return {
+            'text': message.content or '',
+            'tool_calls': tool_calls if tool_calls else None,
+            'tokens_in': response.usage.prompt_tokens,
+            'tokens_out': response.usage.completion_tokens,
+            'stop_reason': choice.finish_reason,
+            'raw_message': message,
+        }
+
+    def append_tool_results(self, messages, ai_response, tool_results):
+        raw_msg = ai_response.get('raw_message')
+        assistant_msg = {'role': 'assistant', 'content': raw_msg.content or ''}
+        if raw_msg.tool_calls:
+            assistant_msg['tool_calls'] = [
+                {
+                    'id': tc.id,
+                    'type': 'function',
+                    'function': {
+                        'name': tc.function.name,
+                        'arguments': tc.function.arguments,
+                    },
+                }
+                for tc in raw_msg.tool_calls
+            ]
+        messages.append(assistant_msg)
+
+        for tr in tool_results:
+            messages.append({
+                'role': 'tool',
+                'tool_call_id': tr['tool_call_id'],
+                'content': tr['result'],
+            })
+
+        return messages

fusion_accounting_ai/services/agent.py

@@ -0,0 +1,947 @@
+import json
+import logging
+import time
+from datetime import timedelta
+
+from odoo import models, fields, api, _
+from odoo.exceptions import UserError
+
+_logger = logging.getLogger(__name__)
+
+# In-memory execution state for live status polling.
+# Key: session_id, Value: {thinking, tool_calls, status}
+# Cleared after each chat() call completes.
+_execution_state = {}
+
+
+def get_execution_state(session_id):
+    """Get the current execution state for a session (called by polling endpoint)."""
+    return _execution_state.get(session_id, {'status': 'idle', 'thinking': '', 'tool_calls': []})
+
+
+# Inter-account transfer pairs: (source_journal, cc_journal, cc_account_pattern)
+# Source sends "MB-CREDIT CARD" (outgoing), CC receives "PAYMENT FROM" (incoming)
+TRANSFER_PAIRS = [
+    # (source_journal_id, cc_journal_id, outstanding_account_id)
+    (50, 51, 493),  # Scotia Current → Passport Visa, Outstanding Receipts - All Banks
+    (53, 28, 493),  # RBC Chequing → RBC Visa, Outstanding Receipts - All Banks
+]
+
+
+class FusionAccountingAgent(models.AbstractModel):
+    _name = 'fusion.accounting.agent'
+    _description = 'Fusion Accounting AI Agent Orchestrator'
+
+    def _get_config(self, key, default=None):
+        ICP = self.env['ir.config_parameter'].sudo()
+        return ICP.get_param(f'fusion_accounting.{key}', default)
+
+    # Domains that need deeper reasoning → use Sonnet
+    COMPLEX_DOMAINS = {'audit', 'month_end', 'hst_management', 'payroll_management'}
+    # Keywords in user messages that suggest complex analysis → use Sonnet
+    COMPLEX_KEYWORDS = {
+        'audit', 'analyze', 'analyse', 'review all', 'full report', 'investigate',
+        'month-end', 'month end', 'close the books', 'hst filing', 'tax return',
+        'what went wrong', 'why is', 'explain the difference', 'compare',
+    }
+
+    def _get_adapter(self):
+        provider = self._get_config('ai_provider', 'claude')
+        if provider == 'claude':
+            return self.env['fusion.accounting.adapter.claude']
+        return self.env['fusion.accounting.adapter.openai']
+
+    def _route_model(self, session, user_message, has_image=False):
+        """Smart model routing: Haiku for routine tool calling, Sonnet for complex analysis.
+        Returns (model_name, can_escalate) — can_escalate=True means Haiku is trying first
+        and we should check if it needs help."""
+        provider = session.ai_provider or self._get_config('ai_provider', 'claude')
+        if provider != 'claude':
+            return None, False
+
+        # Always use Sonnet for images (vision quality matters for OCR)
+        if has_image:
+            return 'claude-sonnet-4-6', False
+
+        # Use Sonnet for complex domains
+        if session.context_domain in self.COMPLEX_DOMAINS:
+            return 'claude-sonnet-4-6', False
+
+        # Use Sonnet if the message contains complex analysis keywords
+        msg_lower = (user_message or '').lower()
+        if any(kw in msg_lower for kw in self.COMPLEX_KEYWORDS):
+            return 'claude-sonnet-4-6', False
+
+        # Default: Haiku with escalation enabled
+        return 'claude-haiku-4-5', True
+
+    def _should_escalate(self, response, tool_calls_log, turn):
+        """Check if Haiku's response suggests it needs Sonnet's help."""
+        text = (response.get('text') or '').lower()
+
+        # Haiku said it can't do something
+        uncertainty_phrases = [
+            "i'm not sure", "i cannot determine", "i don't have enough",
+            "unable to", "i'm unable", "this is complex", "beyond my",
+            "i need more context", "difficult to assess", "i apologize",
+            "i'm having trouble", "let me think about this differently",
+        ]
+        if any(phrase in text for phrase in uncertainty_phrases):
+            return True
+
+        # Haiku made no tool calls on first turn when it probably should have
+        # (user asked a question but Haiku just gave text without using tools)
+        if turn == 0 and not response.get('tool_calls') and not text:
+            return True
+
+        # Haiku had multiple tool errors
+        error_count = sum(1 for tc in tool_calls_log if tc.get('status') == 'error')
+        if error_count >= 2:
+            return True
+
+        # Response is very short for a data question (Haiku might be confused)
+        if turn == 0 and not response.get('tool_calls') and len(text) < 50:
+            return True
+
+        return False
+
+    def _get_tool_registry(self):
+        return self.env['fusion.accounting.tool'].search([('active', '=', True)])
+
+    def _get_tools_for_user(self, user=None):
+        user = user or self.env.user
+        tools = self._get_tool_registry()
+        filtered = self.env['fusion.accounting.tool']
+        for tool in tools:
+            if not tool.required_groups:
+                filtered |= tool
+                continue
+            group_xmlids = [g.strip() for g in tool.required_groups.split(',') if g.strip()]
+            if all(user.has_group(g) for g in group_xmlids):
+                filtered |= tool
+        return filtered
+
+    def _build_tool_definitions(self, tools):
+        definitions = []
+        for tool in tools:
+            # A2: Include tier info in description so AI knows which tools need approval
+            tier_label = {'1': 'Read-only', '2': 'Auto-approved', '3': 'Requires user approval'}.get(tool.tier, '')
+            desc = tool.description or ''
+            if tier_label:
+                desc = f"[Tier {tool.tier}: {tier_label}] {desc}"
+            defn = {
+                'name': tool.name,
+                'description': desc,
+            }
+            if tool.parameters_schema:
+                try:
+                    defn['parameters'] = json.loads(tool.parameters_schema)
+                except json.JSONDecodeError:
+                    defn['parameters'] = {'type': 'object', 'properties': {}}
+            else:
+                defn['parameters'] = {'type': 'object', 'properties': {}}
+            definitions.append(defn)
+        return definitions
+
+    def _load_rules(self, domain=None):
+        rule_domain = [('active', '=', True), ('company_id', '=', self.env.company.id)]
+        if domain:
+            rule_domain.append(('rule_type', '=', domain))
+        rules = self.env['fusion.accounting.rule'].search(rule_domain, order='sequence')
+        admin_rules = rules.filtered(lambda r: r.created_by == 'admin')
+        ai_auto = rules.filtered(lambda r: r.created_by == 'ai' and r.approval_tier == 'auto')
+        ai_pending = rules.filtered(lambda r: r.created_by == 'ai' and r.approval_tier == 'needs_approval')
+        return admin_rules | ai_auto | ai_pending
+
+    def _load_match_history(self, domain=None, limit=None):
+        limit = limit or int(self._get_config('history_in_prompt', '50'))
+        history_domain = [('company_id', '=', self.env.company.id)]
+        if domain:
+            history_domain.append(('tool_name', 'ilike', domain))
+        return self.env['fusion.accounting.match.history'].search(
+            history_domain, limit=limit, order='proposed_at desc',
+        )
+
+    def _build_system_prompt(self, rules, history, context=None, domain=None):
+        from .prompts.system_prompt import build_system_prompt
+        from .prompts.domain_prompts import get_domain_prompt
+        base = build_system_prompt(rules, history, context)
+        if domain:
+            domain_prompt = get_domain_prompt(domain)
+            if domain_prompt:
+                base = f"{base}\n\n{domain_prompt}"
+        return base
+
+    def _execute_tool(self, tool_name, params, session_id=None):
+        from .tools import TOOL_DISPATCH
+        if tool_name not in TOOL_DISPATCH:
+            return {'error': f'Unknown tool: {tool_name}'}
+        try:
+            result = TOOL_DISPATCH[tool_name](self.env, params)
+            return result
+        except Exception as e:
+            _logger.error("Tool execution error (%s): %s", tool_name, e)
+            return {'error': str(e)}
+
+    def _log_match_history(self, session, tool_name, params, result, reasoning='',
+                           confidence=0.0, rule=None, tier='1'):
+        vals = {
+            'session_id': session.id if session else False,
+            'tool_name': tool_name,
+            'tool_params': json.dumps(params, indent=2, default=str) if params else '{}',
+            'tool_result': json.dumps(result, indent=2, default=str) if result else '{}',
+            'ai_reasoning': reasoning,
+            'ai_confidence': confidence,
+            'rule_id': rule.id if rule else False,
+            'proposed_at': fields.Datetime.now(),
+            'decision': 'auto' if tier in ('1', '2') else 'pending',
+            'company_id': self.env.company.id,
+        }
+        return self.env['fusion.accounting.match.history'].sudo().create(vals)
+
+    def chat(self, session_id, user_message, context=None, image=None):
+        session = self.env['fusion.accounting.session'].browse(session_id)
+        if not session.exists():
+            raise UserError(_("Session not found."))
+
+        adapter = self._get_adapter()
+        provider = self._get_config('ai_provider', 'claude')
+
+        # Pin provider to session to prevent cross-adapter message contamination (C5)
+        if session.ai_provider and session.ai_provider != provider:
+            _logger.warning(
+                "Session %s was started with %s but current provider is %s. "
+                "Keeping original provider to avoid message format conflicts.",
+                session.name, session.ai_provider, provider,
+            )
+            provider = session.ai_provider
+            if provider == 'claude':
+                adapter = self.env['fusion.accounting.adapter.claude']
+            else:
+                adapter = self.env['fusion.accounting.adapter.openai']
+
+        tools = self._get_tools_for_user()
+        tool_definitions = self._build_tool_definitions(tools)
+        rules = self._load_rules()
+        history = self._load_match_history()
+        system_prompt = self._build_system_prompt(
+            rules, history, context, domain=session.context_domain,
+        )
+
+        messages_json = json.loads(session.message_ids_json or '[]')
+
+        # Build user message — may include image for vision
+        if image and isinstance(image, dict) and image.get('base64'):
+            user_content = []
+            if user_message:
+                user_content.append({'type': 'text', 'text': user_message})
+            user_content.append({
+                'type': 'image',
+                'source': {
+                    'type': 'base64',
+                    'media_type': image.get('media_type', 'image/png'),
+                    'data': image['base64'],
+                },
+            })
+            messages_json.append({'role': 'user', 'content': user_content})
+        else:
+            messages_json.append({'role': 'user', 'content': user_message})
+
+        # Smart model routing: Haiku for routine, Sonnet for complex
+        has_image = bool(image and isinstance(image, dict) and image.get('base64'))
+        model_override, can_escalate = self._route_model(session, user_message, has_image=has_image)
+        escalated = False
+        if model_override:
+            _logger.info("Model routing: %s → %s (escalation=%s)", session.name, model_override, can_escalate)
+
+        max_turns = max(int(self._get_config('max_tool_calls', '20')), 1)
+        total_tokens_in = 0
+        total_tokens_out = 0
+        response = {'text': '', 'tool_calls': None}
+        has_pending_tier3 = False
+        tool_calls_log = []  # Track tool calls for frontend display
+        reconciliation_data = None  # Raw data from suggest_bank_line_matches
+
+        # Initialize live execution state for polling
+        _execution_state[session.id] = {
+            'status': 'thinking',
+            'thinking': '',
+            'tool_calls': [],
+            'turn': 0,
+        }
+
+        for turn in range(max_turns):
+            _execution_state[session.id]['status'] = 'calling_ai'
+            _execution_state[session.id]['turn'] = turn + 1
+
+            response = adapter.call_with_tools(
+                system_prompt=system_prompt,
+                messages=messages_json,
+                tools=tool_definitions,
+                model_override=model_override,
+            )
+            total_tokens_in += response.get('tokens_in', 0)
+            total_tokens_out += response.get('tokens_out', 0)
+
+            # Check if Haiku needs to escalate to Sonnet
+            if can_escalate and not escalated and self._should_escalate(response, tool_calls_log, turn):
+                _logger.info("Escalating %s from Haiku → Sonnet (turn %d)", session.name, turn)
+                model_override = 'claude-sonnet-4-6'
+                escalated = True
+                can_escalate = False
+                _execution_state[session.id]['status'] = 'escalating'
+                # Re-call with Sonnet
+                response = adapter.call_with_tools(
+                    system_prompt=system_prompt,
+                    messages=messages_json,
+                    tools=tool_definitions,
+                    model_override=model_override,
+                )
+                total_tokens_in += response.get('tokens_in', 0)
+                total_tokens_out += response.get('tokens_out', 0)
+
+            # Capture thinking text for live display
+            thinking = ''
+            for block in (response.get('raw_content') or []):
+                if hasattr(block, 'type') and block.type == 'thinking':
+                    thinking = block.thinking
+                    break
+            if thinking:
+                _execution_state[session.id]['thinking'] = thinking[:500]  # Truncated for live display
+
+            if response.get('tool_calls'):
+                tool_results = []
+                _execution_state[session.id]['status'] = 'calling_tools'
+
+                for tc in response['tool_calls']:
+                    tool_name = tc['name']
+                    tool_params = tc.get('arguments', {})
+                    tool_rec = tools.filtered(lambda t: t.name == tool_name)
+                    tier = tool_rec.tier if tool_rec else '1'
+
+                    # Update live state: show which tool is running
+                    _execution_state[session.id]['tool_calls'].append({
+                        'name': tool_name, 'status': 'running',
+                    })
+
+                    if tier == '3':
+                        has_pending_tier3 = True
+                        history_rec = self._log_match_history(
+                            session, tool_name, tool_params, None,
+                            reasoning=thinking or '',
+                            confidence=tc.get('confidence', 0.0),
+                            tier='3',
+                        )
+                        tool_results.append({
+                            'tool_call_id': tc.get('id', ''),
+                            'result': json.dumps({
+                                'status': 'pending_approval',
+                                'match_history_id': history_rec.id,
+                                'message': f'Action requires user approval. Match history ID: {history_rec.id}',
+                            }),
+                        })
+                        tool_calls_log.append({
+                            'name': tool_name,
+                            'tier': tier,
+                            'status': 'pending_approval',
+                            'summary': self._build_tool_call_summary(tool_name, tool_params, None),
+                        })
+                        _execution_state[session.id]['tool_calls'][-1]['status'] = 'pending'
+                    else:
+                        t0 = time.time()
+                        result = self._execute_tool(tool_name, tool_params, session.id)
+                        elapsed = round((time.time() - t0) * 1000)
+                        self._log_match_history(
+                            session, tool_name, tool_params, result,
+                            reasoning=thinking or '',
+                            tier=tier,
+                        )
+                        tool_results.append({
+                            'tool_call_id': tc.get('id', ''),
+                            'result': json.dumps(result) if not isinstance(result, str) else result,
+                        })
+                        tc_status = 'error' if isinstance(result, dict) and result.get('error') else 'ok'
+                        tc_summary = self._build_tool_call_summary(tool_name, tool_params, result)
+
+                        # Capture reconciliation data for direct frontend rendering
+                        if tool_name == 'suggest_bank_line_matches' and tc_status == 'ok':
+                            reconciliation_data = result
+                        tool_calls_log.append({
+                            'name': tool_name,
+                            'tier': tier,
+                            'status': tc_status,
+                            'summary': tc_summary,
+                            'duration_ms': elapsed,
+                        })
+                        # Update live state
+                        _execution_state[session.id]['tool_calls'][-1].update({
+                            'status': tc_status, 'summary': tc_summary, 'duration_ms': elapsed,
+                        })
+                        try:
+                            self._check_rule_proposal(tool_name, tool_params, session)
+                        except Exception:
+                            _logger.exception("Error in _check_rule_proposal for tool %s", tool_name)
+
+                messages_json = adapter.append_tool_results(
+                    messages_json, response, tool_results,
+                )
+                session.write({'tool_call_count': session.tool_call_count + len(tool_results)})
+
+                # C2: Short-circuit loop when Tier 3 actions are pending —
+                # force a final text response so the AI can present approval cards
+                if has_pending_tier3:
+                    try:
+                        response = adapter.call_with_tools(
+                            system_prompt=system_prompt,
+                            messages=messages_json,
+                            tools=[],
+                            model_override=model_override,
+                        )
+                        total_tokens_in += response.get('tokens_in', 0)
+                        total_tokens_out += response.get('tokens_out', 0)
+                        messages_json.append({
+                            'role': 'assistant',
+                            'content': response.get('text', 'I have proposed actions that require your approval.'),
+                        })
+                    except Exception:
+                        messages_json.append({
+                            'role': 'assistant',
+                            'content': 'I have proposed actions that require your approval. Please review the pending items above.',
+                        })
+                    break
+            else:
+                assistant_text = response.get('text', '')
+                messages_json.append({'role': 'assistant', 'content': assistant_text})
+                break
+        else:
+            # Loop exhausted — force a final text response without tools
+            try:
+                response = adapter.call_with_tools(
+                    system_prompt=system_prompt,
+                    messages=messages_json,
+                    tools=[],
+                    model_override=model_override,
+                )
+                total_tokens_in += response.get('tokens_in', 0)
+                total_tokens_out += response.get('tokens_out', 0)
+                messages_json.append({
+                    'role': 'assistant',
+                    'content': response.get('text', 'I reached the maximum number of steps. Please continue the conversation.'),
+                })
+            except Exception:
+                _logger.warning("Failed to get final summary after max tool calls")
+
+        session.write({
+            'message_ids_json': json.dumps(messages_json),
+            'token_count_in': session.token_count_in + total_tokens_in,
+            'token_count_out': session.token_count_out + total_tokens_out,
+            'ai_provider': provider,
+            'ai_model': model_override or adapter._get_model_name(),
+        })
+
+        pending = self.env['fusion.accounting.match.history'].search([
+            ('session_id', '=', session.id),
+            ('decision', '=', 'pending'),
+        ])
+
+        # Clear live execution state
+        _execution_state.pop(session.id, None)
+
+        # Add escalation marker to tool calls log if it happened
+        if escalated:
+            tool_calls_log.insert(0, {
+                'name': 'model_escalation',
+                'tier': '-',
+                'status': 'ok',
+                'summary': 'Escalated from Haiku to Sonnet for deeper analysis',
+                'duration_ms': 0,
+            })
+
+        result_payload = {
+            'text': response.get('text', ''),
+            'tool_calls_log': tool_calls_log,
+            'pending_approvals': [self._format_pending_approval(p) for p in pending],
+            'session_id': session.id,
+            'model_used': model_override or adapter._get_model_name(),
+        }
+
+        # Attach raw reconciliation data so frontend renders it directly
+        # (instead of relying on AI to format fusion-table JSON correctly)
+        if reconciliation_data:
+            result_payload['reconciliation_table'] = reconciliation_data
+
+        return result_payload
+
+    def _build_tool_call_summary(self, tool_name, params, result):
+        """Build a one-line summary of what a tool call did, for the collapsed tool log."""
+        try:
+            # Result-based summaries (when we have output)
+            if result and isinstance(result, dict) and not result.get('error'):
+                count = result.get('count')
+                status = result.get('status')
+                if status == 'created':
+                    name = result.get('name', '')
+                    return f"Created {name}" if name else "Created successfully"
+                if status == 'matched':
+                    return "Matched successfully"
+                if count is not None:
+                    return f"Found {count} result{'s' if count != 1 else ''}"
+                if 'balance' in result:
+                    return f"Balance: ${result['balance']:,.2f}"
+                if 'total' in result:
+                    return f"Total: ${result['total']:,.2f}"
+                if 'entries' in result:
+                    return f"Found {len(result['entries'])} entries"
+                if 'accounts' in result:
+                    return f"Found {len(result['accounts'])} accounts"
+                if status:
+                    return str(status)
+
+            if result and isinstance(result, dict) and result.get('error'):
+                err = str(result['error'])
+                return f"Error: {err[:80]}"
+
+            # Params-based summaries (for pending approvals, no result yet)
+            if params:
+                ref = params.get('ref', params.get('reference', params.get('name', '')))
+                amount = params.get('amount')
+                lines = params.get('lines', [])
+                if lines:
+                    total = sum(l.get('debit', 0) for l in lines)
+                    return f"{ref} — ${total:,.2f}" if ref else f"${total:,.2f} journal entry"
+                if ref and amount:
+                    return f"{ref} — ${abs(amount):,.2f}"
+                if ref:
+                    return str(ref)
+
+            return "Completed"
+        except Exception:
+            return "Completed"
+
+    def _format_pending_approval(self, history):
+        """Build a rich approval payload so the UI can show exactly what's being approved."""
+        params = {}
+        try:
+            params = json.loads(history.tool_params or '{}')
+        except json.JSONDecodeError:
+            pass
+
+        # Extract amount from params — look in common locations
+        amount = history.amount or 0.0
+        if not amount:
+            # Try to compute from journal entry lines
+            lines = params.get('lines', [])
+            if lines:
+                amount = sum(l.get('debit', 0) for l in lines)
+            # Or from direct amount field
+            if not amount:
+                amount = abs(params.get('amount', 0))
+
+        # Build a human-readable summary of what this action will do
+        summary = self._build_approval_summary(history.tool_name, params)
+
+        return {
+            'id': history.id,
+            'tool_name': history.tool_name,
+            'params': history.tool_params,
+            'reasoning': history.ai_reasoning,
+            'confidence': history.ai_confidence,
+            'amount': amount,
+            'summary': summary,
+        }
+
+    def _resolve_account_label(self, account_id):
+        """Resolve an account ID to 'code - name' for display."""
+        if not account_id:
+            return '?'
+        try:
+            acct = self.env['account.account'].browse(int(account_id))
+            if acct.exists():
+                return f"{acct.code} {acct.name}"
+        except Exception:
+            pass
+        return str(account_id)
+
+    def _build_approval_summary(self, tool_name, params):
+        """Generate a short human-readable description of what a Tier 3 action will do."""
+        try:
+            if tool_name == 'create_payroll_journal_entry':
+                ref = params.get('ref', 'Payroll Entry')
+                date = params.get('date', '?')
+                lines = params.get('lines', [])
+                total = sum(l.get('debit', 0) for l in lines)
+                acct_names = []
+                for l in lines:
+                    aid = l.get('account_id', '')
+                    acct_label = self._resolve_account_label(aid)
+                    if l.get('debit'):
+                        acct_names.append(f"Dr {acct_label}: ${l['debit']:,.2f}")
+                    elif l.get('credit'):
+                        acct_names.append(f"Cr {acct_label}: ${l['credit']:,.2f}")
+                detail = ' / '.join(acct_names) if acct_names else ''
+                return f"{ref} on {date} — ${total:,.2f}\n{detail}"
+
+            elif tool_name == 'create_vendor_bill':
+                partner = params.get('partner_name', params.get('partner_id', '?'))
+                amount = params.get('amount', 0)
+                ref = params.get('ref', params.get('reference', ''))
+                date = params.get('date', '?')
+                return f"Vendor bill for {partner} — ${abs(amount):,.2f} on {date}" + (f" ({ref})" if ref else "")
+
+            elif tool_name == 'register_bill_payment':
+                bill_id = params.get('bill_id', '?')
+                amount = params.get('amount', 0)
+                journal = params.get('journal_id', '?')
+                return f"Pay bill #{bill_id} — ${abs(amount):,.2f} from journal {journal}"
+
+            elif tool_name == 'create_expense_entry':
+                ref = params.get('ref', params.get('memo', 'Expense'))
+                amount = params.get('amount', 0)
+                account = params.get('expense_account_id', '?')
+                return f"{ref} — ${abs(amount):,.2f} to account {account}"
+
+            elif tool_name == 'register_hst_payment':
+                amount = params.get('amount', 0)
+                date = params.get('date', '?')
+                return f"HST remittance — ${abs(amount):,.2f} on {date}"
+
+            elif tool_name in ('apply_payment', 'send_followup', 'create_payment_reminder'):
+                partner = params.get('partner_name', params.get('partner_id', '?'))
+                amount = params.get('amount', 0)
+                return f"{tool_name.replace('_', ' ').title()} for {partner}" + (f" — ${abs(amount):,.2f}" if amount else "")
+
+            elif tool_name == 'flag_entry':
+                move_id = params.get('move_id', '?')
+                reason = params.get('reason', '')
+                return f"Flag entry #{move_id}" + (f": {reason}" if reason else "")
+
+            else:
+                # Generic fallback: show key params
+                parts = []
+                for key in ('ref', 'reference', 'name', 'partner_name', 'date', 'move_id'):
+                    if key in params:
+                        parts.append(f"{key}: {params[key]}")
+                if 'amount' in params:
+                    parts.append(f"${abs(params['amount']):,.2f}")
+                return ' | '.join(parts) if parts else json.dumps(params)[:120]
+
+        except Exception:
+            return str(params)[:120]
+
+    def approve_action(self, match_history_id):
+        history = self.env['fusion.accounting.match.history'].browse(match_history_id)
+        if not history.exists() or history.decision != 'pending':
+            raise UserError(_("Action not found or already decided."))
+
+        params = json.loads(history.tool_params or '{}')
+        result = self._execute_tool(history.tool_name, params, history.session_id.id)
+
+        history.write({
+            'decision': 'approved',
+            'decided_at': fields.Datetime.now(),
+            'decided_by': self.env.user.id,
+            'tool_result': json.dumps(result) if not isinstance(result, str) else result,
+        })
+        if history.rule_id:
+            history.rule_id.sudo()._record_decision(approved=True)
+
+        # C1: Update session messages_json so next chat turn has coherent history
+        self._update_session_after_decision(history, result)
+
+        # M8: Trigger promotion check after approval
+        try:
+            self.env['fusion.accounting.scoring'].check_promotions()
+        except Exception:
+            _logger.exception("Error checking promotions after approval")
+
+        return result
+
+    def _check_rule_proposal(self, tool_name, params, session):
+        """Detect repeated patterns and propose new rules when 3+ identical matches."""
+        recent = self.env['fusion.accounting.match.history'].search([
+            ('tool_name', '=', tool_name),
+            ('decision', 'in', ('approved', 'auto')),
+            ('company_id', '=', self.env.company.id),
+        ], limit=20, order='proposed_at desc')
+
+        if len(recent) < 3:
+            return
+
+        from collections import Counter
+        param_keys = []
+        for h in recent:
+            try:
+                p = json.loads(h.tool_params or '{}')
+                key_parts = []
+                for k in sorted(p.keys()):
+                    if k not in ('limit', 'date_from', 'date_to'):
+                        key_parts.append(f'{k}={json.dumps(p[k], sort_keys=True)}')
+                if key_parts:
+                    param_keys.append('|'.join(key_parts))
+            except json.JSONDecodeError:
+                continue
+
+        counts = Counter(param_keys)
+        for pattern, count in counts.most_common(3):
+            if count < 3:
+                break
+            existing = self.env['fusion.accounting.rule'].search([
+                ('match_logic', 'ilike', pattern[:50]),
+                ('company_id', '=', self.env.company.id),
+            ], limit=1)
+            if existing:
+                continue
+
+            self.env['fusion.accounting.rule'].create({
+                'name': f'AI Pattern: {tool_name} ({pattern[:40]})',
+                'rule_type': 'match',
+                'description': f'Automatically detected pattern from {count} approved uses of {tool_name}.',
+                'match_logic': f'When using {tool_name} with parameters matching: {pattern}',
+                'created_by': 'ai',
+                'approval_tier': 'needs_approval',
+                'company_id': self.env.company.id,
+            })
+            _logger.info("AI proposed rule for pattern: %s (%d matches)", tool_name, count)
+
+    def reject_action(self, match_history_id, reason=''):
+        history = self.env['fusion.accounting.match.history'].browse(match_history_id)
+        if not history.exists() or history.decision != 'pending':
+            raise UserError(_("Action not found or already decided."))
+
+        history.write({
+            'decision': 'rejected',
+            'decided_at': fields.Datetime.now(),
+            'decided_by': self.env.user.id,
+            'rejection_reason': reason,
+        })
+        if history.rule_id:
+            history.rule_id.sudo()._record_decision(approved=False)
+
+        # C1: Update session messages_json so next chat turn has coherent history
+        reject_result = {'status': 'rejected', 'reason': reason}
+        self._update_session_after_decision(history, reject_result)
+
+        return reject_result
+
+    def _update_session_after_decision(self, history, result):
+        """Update session messages_json to replace pending_approval placeholder
+        with actual tool result, preventing dangling tool_use blocks."""
+        session = history.session_id
+        if not session or not session.message_ids_json:
+            return
+        try:
+            messages = json.loads(session.message_ids_json)
+            result_str = json.dumps(result) if not isinstance(result, str) else result
+            updated = False
+            for msg in messages:
+                if msg.get('role') != 'user':
+                    continue
+                content = msg.get('content')
+                if isinstance(content, list):
+                    for block in content:
+                        if (isinstance(block, dict) and block.get('type') == 'tool_result'
+                                and 'pending_approval' in str(block.get('content', ''))):
+                            # Check if this is the matching tool_result block
+                            if str(history.id) in str(block.get('content', '')):
+                                block['content'] = result_str
+                                updated = True
+                                break
+                    if updated:
+                        break
+            if updated:
+                session.write({'message_ids_json': json.dumps(messages)})
+        except Exception:
+            _logger.warning("Failed to update session messages after decision for history %s", history.id)
+
+    # ----------------------------------------------------------------
+    # Cron: Auto-Reconcile Inter-Account Transfers
+    # ----------------------------------------------------------------
+    @api.model
+    def _cron_reconcile_transfers(self):
+        """Automatically reconcile inter-account credit card payments.
+
+        When a payment is made from a bank account (e.g. Scotia Current) to a
+        credit card (e.g. Scotia Passport Visa), two bank statement lines appear:
+          - Source side: "MB-CREDIT CARD" (negative) — reconciled by model 38/35
+          - CC side: "PAYMENT FROM *7814" (positive) — needs matching
+
+        The source-side reconciliation creates outstanding entries on account 493.
+        This cron matches the CC-side lines against those outstanding entries by
+        exact amount and closest date (within 3 days).
+        """
+        AML = self.env['account.move.line'].sudo()
+        BSL = self.env['account.bank.statement.line'].sudo()
+        company_partner_id = self.env.company.partner_id.id
+
+        total_reconciled = 0
+
+        for source_jid, cc_jid, outstanding_acct_id in TRANSFER_PAIRS:
+            # Find all unreconciled INCOMING lines on the credit card journal
+            cc_lines = BSL.search([
+                ('journal_id', '=', cc_jid),
+                ('is_reconciled', '=', False),
+                ('amount', '>', 0),  # Incoming payments only
+                ('company_id', '=', self.env.company.id),
+            ])
+            if not cc_lines:
+                continue
+
+            journal_name = cc_lines[0].journal_id.name
+            _logger.info(
+                "Transfer reconcile: %s — %d incoming unreconciled lines",
+                journal_name, len(cc_lines),
+            )
+
+            reconciled = 0
+            skipped = 0
+
+            for line in cc_lines:
+                line_date = line.move_id.date
+                amount = line.amount
+
+                # Find outstanding entries with exact matching amount
+                candidates = AML.search([
+                    ('account_id', '=', outstanding_acct_id),
+                    ('partner_id', '=', company_partner_id),
+                    ('reconciled', '=', False),
+                    ('amount_residual', '=', amount),
+                ])
+
+                if not candidates:
+                    skipped += 1
+                    continue
+
+                # Pick the candidate closest in date (within 3 days)
+                best = None
+                best_gap = 999
+                for c in candidates:
+                    gap = abs((c.date - line_date).days)
+                    if gap < best_gap:
+                        best_gap = gap
+                        best = c
+
+                if best_gap > 7:
+                    skipped += 1
+                    continue
+
+                # Set partner and reconcile
+                try:
+                    line.partner_id = company_partner_id
+                    line.set_line_bank_statement_line(best.ids)
+                    reconciled += 1
+                except Exception as e:
+                    _logger.warning(
+                        "Transfer reconcile failed: line %s (%s, $%.2f): %s",
+                        line.id, line.payment_ref, amount, e,
+                    )
+
+                # Commit every 50 lines to avoid long transactions
+                if reconciled % 50 == 0 and reconciled > 0:
+                    self.env.cr.commit()
+
+            self.env.cr.commit()
+            total_reconciled += reconciled
+            _logger.info(
+                "Transfer reconcile: %s — reconciled %d, skipped %d",
+                journal_name, reconciled, skipped,
+            )
+
+        _logger.info("Transfer reconcile complete: %d total reconciled", total_reconciled)
+
+    # ----------------------------------------------------------------
+    # One-time: Match payroll cheque bank lines against open payroll liability entries
+    # ----------------------------------------------------------------
+    @api.model
+    def _reconcile_payroll_cheques(self):
+        """Reconcile payroll cheque bank lines using writeoff to Payroll Liabilities (2201).
+
+        Your payroll JEs post:
+            Dr Salaries / Dr ER CPP-EI / Dr CRA Taxes
+                Cr 2201 Payroll Liabilities (net pay = cheque amount)
+
+        When the cheque clears the bank, the bank line shows:
+            "Cheque 1773 : Cheque" = -$1,477.95
+
+        This method finds cheque bank lines that have a matching payroll liability
+        entry (same amount) and applies a reconcile model that writes off to account
+        433 (Payroll Liabilities). This debits 433 to clear the liability.
+
+        Non-payroll cheques (no matching entry on 433) are skipped.
+        """
+        PAYROLL_LIABILITY_ACCT_ID = 433  # code 2201
+        SCOTIA_CURRENT_JOURNAL_ID = 50
+
+        AML = self.env['account.move.line'].sudo()
+        BSL = self.env['account.bank.statement.line'].sudo()
+        RecModel = self.env['account.reconcile.model'].sudo()
+
+        # Find the payroll cheque reconcile model (must be pre-created via XML or manually)
+        model = RecModel.search([
+            ('name', 'ilike', 'Payroll Cheque'),
+            ('company_id', '=', self.env.company.id),
+        ], limit=1)
+
+        if not model:
+            _logger.warning("Payroll cheque reconcile: no 'Payroll Cheque' model found — create one manually")
+            return
+
+        # Find all unreconciled cheque lines on Scotia Current (negative = outgoing)
+        # Only process lines after lock date to avoid lock date errors
+        cheque_lines = BSL.search([
+            ('journal_id', '=', SCOTIA_CURRENT_JOURNAL_ID),
+            ('is_reconciled', '=', False),
+            ('amount', '<', 0),
+            ('payment_ref', 'ilike', 'cheque'),
+            ('company_id', '=', self.env.company.id),
+        ], order='move_id asc')
+
+        # Filter to post-lock-date lines only
+        lock_date = self.env.company.fiscalyear_lock_date
+        if lock_date:
+            cheque_lines = cheque_lines.filtered(lambda l: l.move_id.date > lock_date)
+
+        _logger.info("Payroll cheque reconcile: found %d unreconciled cheque lines (post lock date)", len(cheque_lines))
+
+        # Build set of all known payroll liability credit amounts
+        payroll_credit_amounts = set()
+        for aml in AML.search([
+            ('account_id', '=', PAYROLL_LIABILITY_ACCT_ID),
+            ('parent_state', '=', 'posted'),
+            ('credit', '>', 0),
+        ]):
+            payroll_credit_amounts.add(round(aml.credit, 2))
+
+        # Filter: only reconcile cheques that have a matching payroll liability entry
+        payroll_lines = cheque_lines.filtered(
+            lambda l: round(abs(l.amount), 2) in payroll_credit_amounts
+        )
+
+        _logger.info(
+            "Payroll cheque reconcile: %d payroll, %d non-payroll (skipped)",
+            len(payroll_lines), len(cheque_lines) - len(payroll_lines),
+        )
+
+        if not payroll_lines:
+            _logger.info("Payroll cheque reconcile: nothing to reconcile")
+            return
+
+        # Apply the reconcile model to payroll cheque lines
+        try:
+            model._apply_reconcile_models(payroll_lines)
+            self.env.cr.commit()
+        except Exception as e:
+            _logger.exception("Payroll cheque reconcile batch failed: %s", e)
+            self.env.cr.rollback()
+            return
+
+        # Count results
+        still_unreconciled = payroll_lines.filtered(lambda l: not l.is_reconciled)
+        reconciled = len(payroll_lines) - len(still_unreconciled)
+
+        for line in still_unreconciled[:10]:
+            _logger.info("Payroll cheque still unreconciled: %s $%.2f", line.payment_ref, abs(line.amount))
+
+        _logger.info(
+            "Payroll cheque reconcile complete: %d reconciled, %d still unreconciled",
+            reconciled, len(still_unreconciled),
+        )

fusion_accounting_ai/services/data_adapters/__init__.py

@@ -0,0 +1 @@
+

fusion_accounting_ai/services/prompts/__init__.py

@@ -0,0 +1,2 @@
+from . import system_prompt
+from . import domain_prompts

fusion_accounting_ai/services/prompts/domain_prompts.py

@@ -0,0 +1,237 @@
+DOMAIN_PROMPTS = {
+    'bank_reconciliation': """
+BANK RECONCILIATION CONTEXT:
+You are helping with bank statement reconciliation. Key concepts:
+- Bank statement lines (account.bank.statement.line) represent transactions from the bank feed.
+- Each line needs to be matched to one or more journal items (account.move.line).
+- Matching is done via set_line_bank_statement_line(move_line_ids).
+- Fee differences (e.g., Elavon card processing fees) should be allocated to the fee account.
+- Weekend batches may combine multiple days of card payments.
+- Always verify amounts before proposing a match.
+
+SMART MATCHING WORKFLOW:
+When the user asks to match or reconcile a specific bank line:
+1. Call suggest_bank_line_matches(statement_line_id=X) to find candidate invoices/bills.
+2. Present the results as a reconciliation-mode fusion-table. IMPORTANT: pass the tool
+   result fields DIRECTLY into the fusion-table — do NOT reformat into cells arrays:
+   ```fusion-table
+   {
+     "mode": "reconciliation",
+     "title": "Match: [ref] $[amount]",
+     "source_tool": "suggest_bank_line_matches",
+     "bank_line": <copy bank_line from tool result>,
+     "candidates": <copy candidates array from tool result>,
+     "best_combination": <copy best_combination from tool result>
+   }
+   ```
+   Each candidate must have: aml_id, name, ref, partner, date, amount_residual, type, score, reasons.
+   Do NOT convert candidates into {"id":..., "cells":[...]} format — use the raw tool output.
+3. The user can: check/uncheck rows, edit amounts for partial payments,
+   search for additional entries via the search bar, then click Apply Match.
+4. When the user clicks Apply Match, you receive a [TABLE_ACTION] with
+   action=apply_match containing AML IDs and custom amounts.
+5. Call match_bank_line_to_payments with the AML IDs from the action
+   (full matches first, partial last — Odoo handles partial on last AML).
+6. Partial payment: if apply_amount < amount_residual, it's partial.
+   Only ONE AML can be partial (the last one). Odoo leaves the residual open.
+
+Bank journal IDs: RBC Chequing=53, Scotia Current=50, Scotia Visa=51, RBC Visa=28.
+""",
+
+    'hst_management': """
+HST/GST MANAGEMENT CONTEXT:
+You are helping with Canadian HST/GST tax management.
+- HST Collected is tracked on account 2005 (credit balance = liability).
+- Input Tax Credits (ITCs) are on account 2006 (debit balance = asset).
+- Net HST = Collected - ITCs. Positive means owing to CRA.
+- Quarterly filing periods. Check for missing tax on invoices/bills.
+- All vendor bills should have ITCs unless explicitly exempt.
+- HST Purchase tax ID is 20 (13%). No Tax Purchase ID is 32 (0%).
+
+HST FILING WORKFLOW (4 phases — follow this order):
+
+PHASE 1 — REPORTS: Run all at once:
+  calculate_hst_balance, get_tax_report, find_missing_itc_bills,
+  find_missing_tax_invoices, audit_tax_compliance.
+  Present summary with HST position (owing vs refund).
+
+PHASE 2 — BANK SWEEP: Check ALL bank accounts for unreconciled expenses:
+  Call get_unreconciled_bank_lines for each bank journal (RBC Chequing 9595=53,
+  Current Account Scotia=50, Scotiabank Passport Visa 8046=51, RBC Visa X 6752=28).
+  Present ALL unreconciled expense lines (negative amounts) as a fusion-table
+  with your recommendation per row.
+
+PHASE 3 — PER-LINE PROCESSING: For each flagged expense line:
+  0. FIRST: check_recurring_pattern(line_id=X) — if match found, follow action_note
+     instructions EXACTLY (account, HST, partner, reconcile model). No user input needed
+     for recurring payments. If a reconcile_model_id is returned, use apply_reconcile_model.
+  1. get_bank_line_details — check if a vendor bill already exists for same amount/date
+  2. find_similar_bank_lines — check history AND vendor_tax_pattern for coding/tax pattern
+  3. CRITICAL: Check vendor_tax_pattern.is_po_vendor flag:
+     - If is_po_vendor=true: This vendor's bills come from Purchase Orders. Do NOT create
+       a new bill. Instead, use get_unpaid_bills to find the existing bill and propose
+       match_bank_line_to_payments to match the bank payment to that bill.
+     - If is_po_vendor=false: Proceed with bill creation workflow below.
+  4. If bill already exists → propose match_bank_line_to_payments
+  5. If no bill but history match → propose create_vendor_bill with same coding pattern
+  6. If no bill and no history → ask user: "Does this expense include HST?"
+  7. search_partners — find the vendor by keyword from the bank description
+  8. Once confirmed → create_vendor_bill + register_bill_payment (Tier 3, needs approval)
+  9. Alternative: user can choose "Direct GL" → create_expense_entry (Tier 3)
+  For expenses that obviously have no HST (bank fees, interest charges, insurance),
+  proactively recommend "No HST" and explain why.
+
+PO-TRACKED VENDORS (do NOT create bills for these — bills come from Purchase Orders):
+  When find_similar_bank_lines returns is_po_vendor=true or the vendor_tax_pattern
+  note starts with "PO-TRACKED VENDOR", the bill already exists or will be created
+  from a PO. Your job is ONLY to find the existing unpaid bill and match the bank
+  payment to it. If no unpaid bill exists, flag it for the user: "This is a PO vendor
+  but no matching bill was found — the PO may not have been billed yet."
+
+PHASE 4 — VERIFICATION: Re-run calculate_hst_balance and get_tax_report
+  to show the updated HST position after all expenses are recorded.
+
+BANK JOURNAL IDS: RBC Chequing 9595=53, Current Account Scotia=50,
+  Scotiabank Passport Visa 8046=51, RBC Visa X 6752=28.
+MISC JOURNAL: ID=3 (for direct GL expense entries).
+""",
+
+    'accounts_receivable': """
+ACCOUNTS RECEIVABLE CONTEXT:
+- AR aging: current, 1-30, 31-60, 61-90, 90+ days overdue.
+- Follow-up actions escalate by aging bucket.
+- Payments should be matched to specific invoices.
+- Unmatched payments sit on the Outstanding Receipts account (1122).
+""",
+
+    'accounts_payable': """
+ACCOUNTS PAYABLE CONTEXT:
+- AP aging mirrors AR: current through 90+ days.
+- Watch for duplicate bills (same vendor + amount + date).
+- Bills should match purchase orders when applicable.
+- Tax on bills should match the vendor's fiscal position.
+""",
+
+    'journal_review': """
+JOURNAL REVIEW CONTEXT:
+- Check for wrong-direction balances (e.g., expense account with credit balance).
+- Detect duplicate entries (same partner + amount + date + journal).
+- Flag entries on unlikely accounts (revenue on a tax account, etc.).
+- Sequence gaps may indicate deleted entries.
+- Draft entries older than 30 days should be reviewed.
+""",
+
+    'month_end': """
+MONTH-END CLOSE CONTEXT:
+- Aggregate all domain checks into a close checklist.
+- Verify all bank reconciliations are current.
+- Check accrual account balances (vacation, sick leave, etc.).
+- Verify no entries exist after lock date.
+- Run hash integrity check.
+- Produce period trial balance summary.
+""",
+
+    'payroll_verification': """
+PAYROLL VERIFICATION CONTEXT:
+- Cross-reference payroll journal entries to bank statement cheques.
+- Verify CPP, EI, and income tax deductions against CRA rate tables.
+- Check CRA remittance account balance vs payments made.
+""",
+
+    'inventory': """
+INVENTORY & COGS CONTEXT:
+- Stock In Hand tracked on account 1069.
+- Price differences on account 5010 (PO price vs bill price).
+- COGS ratio by product category helps spot anomalies.
+- Large inventory adjustments need review.
+""",
+
+    'adp': """
+ADP (ASSISTIVE DEVICE PROGRAM) RECONCILIATION CONTEXT:
+- ADP Receivable tracked on account 1101.
+- ADP invoices have customer portion + ADP portion = total.
+- Government deposits arrive on Scotia Current (journal 50) with label "Assistive Devices : Miscellaneous Payment".
+- ADP partner in Odoo: "ADP (Assistive Device Program)" (id 3421).
+
+ADP PAYMENT MATCHING WORKFLOW:
+1. When user says "match ADP payment" or "check ADP payments":
+   - Call get_unreconciled_bank_lines(journal_id=50) and filter for "Assistive Devices" lines.
+   - For each ADP bank line, call suggest_bank_line_matches(statement_line_id=X).
+   - The tool finds outstanding payments (PBNK2 entries on account 1050) for the ADP partner.
+   - Present as reconciliation fusion-table.
+
+2. When user uploads an ADP remittance advice image:
+   - Read the image. It is a table with these columns:
+     Invoice Number | Invoice Date | Claim Number | Client Ref | Payment Date | Payment Amount
+   - The bottom shows "Total Payment Due: $XX,XXX.XX" — this is the bank deposit amount.
+   - Extract every row: invoice number and payment amount.
+   - Find the bank line on Scotia Current matching the total amount.
+   - Call suggest_bank_line_matches for that bank line.
+   - The outstanding payments on 1050 should sum to the total.
+
+3. When matching, outstanding payments (PBNK2 entries) are preferred over raw invoices.
+   Each PBNK2 entry represents a registered payment batch. Two or more PBNK2 entries
+   may combine to equal the bank deposit total.
+""",
+
+    'reporting': """
+FINANCIAL REPORTING CONTEXT:
+- Reports use Odoo's account.report engine.
+- Available: P&L, Balance Sheet, Trial Balance, Cash Flow.
+- Period comparison available for trend analysis.
+- Export to PDF or XLSX for external distribution.
+""",
+
+    'audit': """
+AUDIT & INTEGRITY CONTEXT:
+- Run comprehensive checks on posted entries.
+- Verify hash chain integrity on journals.
+- Check sequence continuity.
+- Flag entries with chatter messages for review tracking.
+- Audit status per account: todo / reviewed / supervised / anomaly.
+""",
+
+    'payroll_management': """
+PAYROLL MANAGEMENT CONTEXT:
+- Parse pasted payroll summaries from QBO or fusion_payroll.
+- Create payroll journal entries with proper debit/credit lines.
+- Match payroll cheques to bank statement lines.
+- Calculate CRA obligations (CPP employer + employee, EI, income tax).
+- Prepare CRA remittance payment entries.
+""",
+}
+
+
+# A3/A5: Aliases so common domain variations still match a prompt
+DOMAIN_ALIASES = {
+    'bank': 'bank_reconciliation',
+    'bank_recon': 'bank_reconciliation',
+    'hst': 'hst_management',
+    'gst': 'hst_management',
+    'tax': 'hst_management',
+    'ar': 'accounts_receivable',
+    'receivable': 'accounts_receivable',
+    'ap': 'accounts_payable',
+    'payable': 'accounts_payable',
+    'journal': 'journal_review',
+    'close': 'month_end',
+    'month_end_close': 'month_end',
+    'payroll': 'payroll_management',
+    'payroll_verify': 'payroll_verification',
+    'stock': 'inventory',
+    'cogs': 'inventory',
+    'report': 'reporting',
+    'reports': 'reporting',
+    'financial': 'reporting',
+}
+
+
+def get_domain_prompt(domain):
+    if not domain:
+        return ''
+    # Try exact match first, then aliases
+    prompt = DOMAIN_PROMPTS.get(domain, '')
+    if not prompt:
+        resolved = DOMAIN_ALIASES.get(domain, domain)
+        prompt = DOMAIN_PROMPTS.get(resolved, '')
+    return prompt

fusion_accounting_ai/services/prompts/system_prompt.py

@@ -0,0 +1,188 @@
+import json
+
+
+def build_system_prompt(rules, history, context=None):
+    parts = [
+        CORE_SYSTEM_PROMPT,
+        _build_rules_section(rules),
+        _build_history_section(history),
+    ]
+    if context:
+        parts.append(_build_context_section(context))
+    return '\n\n'.join(p for p in parts if p)
+
+
+CORE_SYSTEM_PROMPT = """You are Fusion AI, an expert accounting co-pilot embedded in Odoo 19.
+You assist with bank reconciliation, HST/GST management, AR/AP analysis, journal review,
+month-end close, payroll, inventory, ADP reconciliation, financial reporting, and auditing.
+
+BEHAVIOUR:
+- Use tools to query and act on Odoo data. Never invent financial figures.
+- For Tier 1 tools: execute immediately and report results.
+- For Tier 2 tools: execute and log. Inform the user what was done.
+- For Tier 3 tools: propose the action with clear reasoning. The user must approve.
+- When proposing a Tier 3 action, explain: what you want to do, why, the amounts involved, and your confidence level.
+- Apply Fusion Rules (below) before general reasoning.
+- Reference match history for patterns the user has approved/rejected before.
+- Use Canadian English. Format monetary amounts with $ and two decimals.
+- When you encounter ambiguity, ask clarifying questions rather than guessing.
+
+RESPONSE FORMATTING:
+- Use rich Markdown formatting in your responses. The chat renders Markdown as HTML.
+- Use **bold** for account names, amounts, and key terms.
+- Use ## and ### headers to organize sections in longer responses.
+- Use bullet lists (- item) for findings, issues, and action items.
+- Use numbered lists (1. item) for sequential steps or ranked items.
+- Use `code` for account codes, reference numbers, and technical IDs.
+- Use --- horizontal rules to separate sections in long reports.
+
+INTERACTIVE TABLES (fusion-table) — MANDATORY FOR ACTIONABLE DATA:
+IMPORTANT: When a tool returns a list of records that the user could act on, you MUST use
+a ```fusion-table block instead of a Markdown table. This is REQUIRED — never use plain
+Markdown tables for actionable data. The fusion-table renders an interactive widget with
+checkboxes, your AI recommendations per row, user input fields, and bulk action buttons.
+
+YOU MUST USE fusion-table FOR: missing ITCs/tax (find_missing_itc_bills, find_missing_tax_invoices),
+duplicate entries (find_duplicate_bills, find_duplicate_entries), overdue invoices (get_overdue_invoices),
+unreconciled lines (get_unreconciled_bank_lines, get_unreconciled_receipts, get_unmatched_payments,
+find_unreconciled_suspense), draft entries (find_draft_entries), wrong balances
+(find_wrong_direction_balances), sequence gaps (find_sequence_gaps), wrong accounts
+(find_wrong_account_entries), unpaid bills (get_unpaid_bills), and any other list where
+the user needs to review, dismiss, flag, or create rules for individual rows.
+
+USE REGULAR MARKDOWN TABLES ONLY FOR: P&L (get_profit_loss), balance sheet (get_balance_sheet),
+trial balance (get_trial_balance), cash flow (get_cash_flow), period summaries, tax reports,
+and any purely informational/read-only data where there is nothing to act on per row.
+
+Format: wrap a JSON object in a ```fusion-table fenced code block:
+
+```fusion-table
+{
+  "mode": "interactive",
+  "title": "Descriptive Title",
+  "columns": ["Col1", "Col2", "Col3"],
+  "rows": [
+    {"id": 123, "cells": ["val1", "val2", "val3"], "recommendation": {"action": "dismiss", "reason": "Brief explanation"}},
+    {"id": 456, "cells": ["val1", "val2", "val3"], "recommendation": {"action": "flag", "reason": "Brief explanation"}}
+  ],
+  "actions": ["dismiss", "flag", "create_rule"],
+  "source_tool": "tool_name_that_produced_this"
+}
+```
+
+- "mode": "interactive" (actionable) or "readonly" (informational but structured)
+- "id": the Odoo record ID (account.move id, account.bank.statement.line id, etc.)
+- "recommendation.action": one of "dismiss", "flag", "create_rule"
+- "recommendation.reason": short explanation of why you recommend this action
+- "actions": which bulk action buttons to show
+- "source_tool": the tool name that produced the data
+- You MUST provide a recommendation for each row when using interactive mode.
+- Format monetary amounts as "$X,XXX.XX" in cells.
+- Always include the record ID so actions can target the correct Odoo record.
+- Add a brief text summary before or after the fusion-table block for context.
+
+LINKING TO ODOO RECORDS:
+- When referencing specific records, include clickable Odoo links.
+- Journal entries: [INV/2026/00123](/odoo/accounting/123) where 123 is the move ID.
+- Partners: [Customer Name](/odoo/contacts/456) where 456 is the partner ID.
+- Accounts: reference by code in bold, e.g. **1001 - Cash**.
+- Bank statement lines: mention the date, reference, and amount clearly.
+- When tool results include record IDs, always link them.
+
+BANK LINE MATCHING:
+When the user asks to match, reconcile, or find matches for a specific bank statement line:
+- ALWAYS use suggest_bank_line_matches(statement_line_id=X) as your PRIMARY tool.
+- It searches outstanding payments FIRST (registered payments on 1050/1051 accounts),
+  then open invoices/bills. Outstanding payments are the correct match — not raw invoices.
+- Present results as a reconciliation-mode fusion-table (mode: "reconciliation").
+- Do NOT manually search for invoices or use find_adp_without_payment for matching.
+- The tool handles partner detection, scoring, and subset-sum automatically.
+- For ADP: bank lines say "Assistive Devices" — the tool maps this to the ADP partner.
+
+ADP (ASSISTIVE DEVICE PROGRAM) WORKFLOW:
+ADP sends batch payments covering multiple customer invoices. The bank deposit label is
+"Assistive Devices : Miscellaneous Payment". The user may upload a screenshot of the
+ADP remittance advice to help match invoices.
+
+When handling ADP payments:
+1. First call suggest_bank_line_matches(statement_line_id=X) — it will find outstanding
+   payments on account 1050 that match the bank amount. These are the registered payments
+   (PBNK2/xxxx/xxxxx entries) that were created when invoices were paid in Odoo.
+2. Present results as a reconciliation fusion-table showing the outstanding payments.
+3. The user may need to combine 2-3 outstanding payments to match the bank deposit total.
+
+When the user attaches an ADP remittance advice image:
+- The image is a table with columns: Invoice Number | Invoice Date | Claim Number |
+  Client Ref | Payment Date | Payment Amount
+- The last row shows "Total Payment Due" with the grand total.
+- Extract ALL invoice numbers and their payment amounts from the image.
+- Present a summary table of what you extracted for confirmation.
+- If the user says "mark these paid" or "register these payments":
+  Call register_adp_batch_payment with the extracted invoices and payment date.
+  This registers each payment and creates outstanding receipts on account 1050.
+  Then find the matching bank deposit and use suggest_bank_line_matches to reconcile.
+- If the user says "match these" or "find the bank deposit":
+  Find the bank line matching the total, call suggest_bank_line_matches.
+
+IMAGE ANALYSIS:
+When the user attaches an image to their message, you can see it directly (vision).
+- Read all text, numbers, and tables from the image.
+- For financial documents: extract invoice numbers, amounts, dates, partner names.
+- For remittance advices: extract the line items and grand total.
+- Always confirm what you extracted before taking action.
+
+TOOL CALLING:
+- Call tools by name with the required parameters.
+- You may call multiple tools in sequence to gather data before proposing an action.
+- Do not exceed the maximum tool calls per turn.
+- When presenting tool results, format them richly with tables, bold amounts, and links.
+"""
+
+
+def _build_rules_section(rules):
+    if not rules:
+        return ''
+    lines = ['ACTIVE FUSION RULES:']
+    for rule in rules:
+        priority = 'ADMIN' if rule.created_by == 'admin' else 'AI'
+        tier = 'auto' if rule.approval_tier == 'auto' else 'needs-approval'
+        conf_str = f', confidence={rule.confidence_score:.0%}, uses={rule.total_uses}' if rule.total_uses > 0 else ''
+        lines.append(
+            f'- [{priority}/{tier}{conf_str}] {rule.name} ({rule.rule_type}): '
+            f'{rule.description or rule.match_logic or "No description"}'
+        )
+        if rule.match_logic:
+            logic_text = rule.match_logic[:500]  # Prevent prompt bloat
+            lines.append(f'  Match logic: {logic_text}')
+    return '\n'.join(lines)
+
+
+def _build_history_section(history):
+    if not history:
+        return ''
+    lines = ['RECENT MATCH HISTORY (learn from these patterns):']
+    # A4: Don't hard-cap at 50 — the caller (_load_match_history) already
+    # respects the history_in_prompt config setting
+    for h in history:
+        status = h.decision
+        reason = ''
+        if h.rejection_reason:
+            reason = f' (reason: {h.rejection_reason})'
+        lines.append(
+            f'- {h.tool_name}: {status}{reason} '
+            f'[confidence={h.ai_confidence:.0%}]'
+        )
+        if h.ai_reasoning:
+            lines.append(f'  Reasoning: {h.ai_reasoning[:200]}')
+    return '\n'.join(lines)
+
+
+def _build_context_section(context):
+    if not context:
+        return ''
+    if isinstance(context, dict):
+        parts = ['CURRENT CONTEXT:']
+        for k, v in context.items():
+            parts.append(f'- {k}: {v}')
+        return '\n'.join(parts)
+    return f'CURRENT CONTEXT: {context}'

fusion_accounting_ai/services/scoring.py

@@ -0,0 +1,61 @@
+import logging
+
+from odoo import models, api
+
+_logger = logging.getLogger(__name__)
+
+
+class FusionAccountingScoring(models.AbstractModel):
+    _name = 'fusion.accounting.scoring'
+    _description = 'Fusion Accounting Confidence Scoring'
+
+    def calculate_confidence(self, tool_name, scenario_key=None):
+        domain = [('tool_name', '=', tool_name)]
+        if scenario_key:
+            domain.append(('tool_params', 'ilike', scenario_key))
+        history = self.env['fusion.accounting.match.history'].search(domain)
+        if not history:
+            return 0.0
+        decided = history.filtered(lambda h: h.decision in ('approved', 'rejected'))
+        if not decided:
+            return 0.0
+        approved = len(decided.filtered(lambda h: h.decision == 'approved'))
+        return approved / len(decided)
+
+    def check_promotions(self):
+        ICP = self.env['ir.config_parameter'].sudo()
+        threshold = float(ICP.get_param('fusion_accounting.tier3_threshold', '0.95'))
+        min_sample = int(ICP.get_param('fusion_accounting.tier3_min_sample', '30'))
+
+        rules = self.env['fusion.accounting.rule'].search([
+            ('active', '=', True),
+            ('approval_tier', '=', 'needs_approval'),
+        ])
+        promoted = self.env['fusion.accounting.rule']
+        for rule in rules:
+            if rule.total_uses >= min_sample and rule.confidence_score >= threshold:
+                rule.approval_tier = 'auto'
+                promoted |= rule
+                _logger.info(
+                    "Promoted rule '%s' to auto (confidence=%.2f, sample=%d)",
+                    rule.name, rule.confidence_score, rule.total_uses,
+                )
+        return promoted
+
+    def get_tool_stats(self, tool_name=None):
+        domain = []
+        if tool_name:
+            domain.append(('tool_name', '=', tool_name))
+        history = self.env['fusion.accounting.match.history'].search(domain)
+
+        stats = {}
+        for h in history:
+            if h.tool_name not in stats:
+                stats[h.tool_name] = {
+                    'total': 0, 'approved': 0, 'rejected': 0,
+                    'pending': 0, 'auto': 0,
+                }
+            stats[h.tool_name]['total'] += 1
+            if h.decision in stats[h.tool_name]:
+                stats[h.tool_name][h.decision] += 1
+        return stats

fusion_accounting_ai/services/tools/__init__.py

@@ -0,0 +1,19 @@
+from .bank_reconciliation import TOOLS as BANK_RECON_TOOLS
+from .hst_management import TOOLS as HST_TOOLS
+from .accounts_receivable import TOOLS as AR_TOOLS
+from .accounts_payable import TOOLS as AP_TOOLS
+from .journal_review import TOOLS as JOURNAL_TOOLS
+from .month_end import TOOLS as MONTH_END_TOOLS
+from .payroll import TOOLS as PAYROLL_TOOLS
+from .inventory import TOOLS as INVENTORY_TOOLS
+from .adp import TOOLS as ADP_TOOLS
+from .reporting import TOOLS as REPORTING_TOOLS
+from .audit import TOOLS as AUDIT_TOOLS
+
+TOOL_DISPATCH = {}
+for tools_dict in [
+    BANK_RECON_TOOLS, HST_TOOLS, AR_TOOLS, AP_TOOLS, JOURNAL_TOOLS,
+    MONTH_END_TOOLS, PAYROLL_TOOLS, INVENTORY_TOOLS, ADP_TOOLS,
+    REPORTING_TOOLS, AUDIT_TOOLS,
+]:
+    TOOL_DISPATCH.update(tools_dict)

fusion_accounting_ai/services/tools/accounts_payable.py

@@ -0,0 +1,406 @@
+import logging
+from odoo import fields
+from datetime import timedelta
+
+_logger = logging.getLogger(__name__)
+
+
+def get_ap_aging(env, params):
+    today = fields.Date.today()
+    domain = [
+        ('account_id.account_type', '=', 'liability_payable'),
+        ('parent_state', '=', 'posted'),
+        ('reconciled', '=', False),
+        ('company_id', '=', env.company.id),
+    ]
+    amls = env['account.move.line'].search(domain)
+
+    buckets = {'current': 0, '1_30': 0, '31_60': 0, '61_90': 0, '90_plus': 0}
+    for aml in amls:
+        amt = abs(aml.amount_residual)
+        if not aml.date_maturity or aml.date_maturity >= today:
+            buckets['current'] += amt
+        else:
+            days = (today - aml.date_maturity).days
+            if days <= 30:
+                buckets['1_30'] += amt
+            elif days <= 60:
+                buckets['31_60'] += amt
+            elif days <= 90:
+                buckets['61_90'] += amt
+            else:
+                buckets['90_plus'] += amt
+
+    return {'total': sum(buckets.values()), 'buckets': buckets, 'line_count': len(amls)}
+
+
+def find_duplicate_bills(env, params):
+    window_days = int(params.get('window_days', 7))
+    bills = env['account.move'].search([
+        ('move_type', 'in', ('in_invoice', 'in_refund')),
+        ('state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ], order='partner_id, amount_total, date')
+
+    duplicates = []
+    prev = None
+    for bill in bills:
+        if prev and (
+            prev.partner_id == bill.partner_id
+            and abs(prev.amount_total - bill.amount_total) < 0.01
+            and abs((prev.date - bill.date).days) <= window_days
+        ):
+            duplicates.append({
+                'bill_1': {'id': prev.id, 'name': prev.name, 'date': str(prev.date)},
+                'bill_2': {'id': bill.id, 'name': bill.name, 'date': str(bill.date)},
+                'partner': bill.partner_id.name,
+                'amount': bill.amount_total,
+            })
+        prev = bill
+
+    return {'count': len(duplicates), 'duplicates': duplicates[:20]}
+
+
+def match_bill_to_po(env, params):
+    bill_id = int(params['bill_id'])
+    bill = env['account.move'].browse(bill_id)
+    if not bill.exists():
+        return {'error': 'Bill not found'}
+    matches = []
+    for line in bill.invoice_line_ids:
+        if line.purchase_line_id:
+            matches.append({
+                'bill_line': line.name or '',
+                'po': line.purchase_line_id.order_id.name,
+                'po_line': line.purchase_line_id.name,
+                'po_qty': line.purchase_line_id.product_qty,
+                'bill_qty': line.quantity,
+                'match': abs(line.quantity - line.purchase_line_id.product_qty) < 0.01,
+            })
+    return {'bill': bill.name, 'matches': matches, 'unmatched_lines': len(bill.invoice_line_ids) - len(matches)}
+
+
+def get_unpaid_bills(env, params):
+    domain = [
+        ('move_type', 'in', ('in_invoice', 'in_refund')),
+        ('state', '=', 'posted'),
+        ('payment_state', 'in', ('not_paid', 'partial')),
+        ('company_id', '=', env.company.id),
+    ]
+    if params.get('partner_id'):
+        domain.append(('partner_id', '=', int(params['partner_id'])))
+    bills = env['account.move'].search(domain, order='invoice_date_due asc', limit=int(params.get('limit', 50)))
+    return {
+        'count': len(bills),
+        'total': sum(b.amount_residual for b in bills),
+        'bills': [{
+            'id': b.id, 'name': b.name,
+            'partner': b.partner_id.name if b.partner_id else '',
+            'amount_total': b.amount_total,
+            'amount_residual': b.amount_residual,
+            'date_due': str(b.invoice_date_due) if b.invoice_date_due else '',
+        } for b in bills],
+    }
+
+
+def verify_bill_taxes(env, params):
+    bill_id = int(params['bill_id'])
+    bill = env['account.move'].browse(bill_id)
+    if not bill.exists():
+        return {'error': 'Bill not found'}
+    issues = []
+    for line in bill.invoice_line_ids:
+        if line.product_id and not line.tax_ids:
+            issues.append({
+                'line': line.name or line.product_id.name,
+                'issue': 'No tax applied to product line',
+            })
+    return {'bill': bill.name, 'issues': issues, 'clean': len(issues) == 0}
+
+
+def get_payment_schedule(env, params):
+    days_ahead = int(params.get('days_ahead', 30))
+    cutoff = fields.Date.today() + timedelta(days=days_ahead)
+    bills = env['account.move'].search([
+        ('move_type', '=', 'in_invoice'),
+        ('state', '=', 'posted'),
+        ('payment_state', 'in', ('not_paid', 'partial')),
+        ('invoice_date_due', '<=', cutoff),
+        ('company_id', '=', env.company.id),
+    ], order='invoice_date_due asc')
+    return {
+        'period': f'Next {days_ahead} days',
+        'total': sum(b.amount_residual for b in bills),
+        'bills': [{
+            'id': b.id, 'name': b.name,
+            'partner': b.partner_id.name if b.partner_id else '',
+            'amount_residual': b.amount_residual,
+            'date_due': str(b.invoice_date_due) if b.invoice_date_due else '',
+        } for b in bills[:50]],
+    }
+
+
+def search_partners(env, params):
+    """Search for partners/vendors by name keyword."""
+    keyword = params.get('keyword', '')
+    if not keyword or len(keyword) < 2:
+        return {'error': 'Keyword must be at least 2 characters'}
+    domain = [('name', 'ilike', keyword), ('company_id', 'in', [env.company.id, False])]
+    if params.get('supplier_only'):
+        domain.append(('supplier_rank', '>', 0))
+    partners = env['res.partner'].search(domain, limit=int(params.get('limit', 20)))
+    return {
+        'count': len(partners),
+        'partners': [{
+            'id': p.id,
+            'name': p.name,
+            'supplier_rank': p.supplier_rank,
+            'customer_rank': p.customer_rank,
+            'vat': p.vat or '',
+            'email': p.email or '',
+            'phone': p.phone or '',
+        } for p in partners],
+    }
+
+
+def find_similar_bank_lines(env, params):
+    """Find past reconciled bank lines with similar description to suggest coding patterns.
+    Also checks vendor bill tax patterns if a partner is identified."""
+    keyword = params.get('keyword', '')
+    if not keyword or len(keyword) < 3:
+        return {'error': 'Keyword must be at least 3 characters'}
+    # Find reconciled bank lines with matching payment_ref
+    lines = env['account.bank.statement.line'].search([
+        ('is_reconciled', '=', True),
+        ('payment_ref', 'ilike', keyword),
+        ('company_id', '=', env.company.id),
+    ], order='date desc', limit=int(params.get('limit', 10)))
+
+    matches = []
+    found_partner_id = None
+    for line in lines:
+        move = line.move_id
+        if not move:
+            continue
+        expense_info = {'account_code': '', 'account_name': '', 'tax_applied': False, 'tax_amount': 0.0}
+        for ml in move.line_ids:
+            if ml.account_id.account_type in ('expense', 'expense_direct_cost', 'expense_depreciation'):
+                expense_info['account_code'] = ml.account_id.code
+                expense_info['account_name'] = ml.account_id.name
+                expense_info['tax_applied'] = bool(ml.tax_ids)
+                expense_info['tax_amount'] = sum(t.amount for t in ml.tax_ids) if ml.tax_ids else 0.0
+                break
+        if line.partner_id and not found_partner_id:
+            found_partner_id = line.partner_id.id
+        matches.append({
+            'id': line.id,
+            'date': str(line.date),
+            'payment_ref': line.payment_ref or '',
+            'amount': line.amount,
+            'partner': line.partner_id.name if line.partner_id else '',
+            'partner_id': line.partner_id.id if line.partner_id else None,
+            'expense_account': expense_info['account_code'],
+            'expense_account_name': expense_info['account_name'],
+            'tax_applied': expense_info['tax_applied'],
+            'tax_rate': expense_info['tax_amount'],
+        })
+
+    result = {
+        'keyword': keyword,
+        'count': len(matches),
+        'matches': matches,
+        'suggestion': matches[0] if matches else None,
+    }
+
+    # Check vendor tax profile cache first (fast), fall back to live query
+    partner_id = found_partner_id or (int(params['partner_id']) if params.get('partner_id') else None)
+    if partner_id:
+        profile = env['fusion.vendor.tax.profile'].search([
+            ('partner_id', '=', partner_id),
+            ('company_id', '=', env.company.id),
+        ], limit=1)
+        if profile:
+            result['vendor_tax_pattern'] = {
+                'source': 'cached_profile',
+                'total_bills': profile.total_bills,
+                'bills_with_tax': profile.bills_with_hst,
+                'bills_no_tax': profile.bills_zero_rated,
+                'avg_tax_pct': profile.avg_tax_pct,
+                'tax_classification': profile.tax_classification,
+                'tax_note': profile.tax_note,
+                'primary_account_id': profile.primary_account_id.id if profile.primary_account_id else None,
+                'primary_account_code': profile.primary_account_code or '',
+                'is_foreign': profile.is_foreign,
+                'is_po_vendor': profile.is_po_vendor,
+                'po_count': profile.po_count,
+            }
+        else:
+            # No cached profile — live query for new/small vendors
+            bills = env['account.move'].search([
+                ('move_type', '=', 'in_invoice'), ('state', '=', 'posted'),
+                ('partner_id', '=', partner_id),
+            ], order='date desc', limit=10)
+            tax_stats = {'source': 'live_query', 'total_bills': len(bills),
+                         'bills_with_tax': 0, 'bills_no_tax': 0,
+                         'avg_tax_pct': 0.0, 'tax_note': ''}
+            tax_pcts = []
+            for bill in bills:
+                if bill.amount_tax > 0.01:
+                    tax_stats['bills_with_tax'] += 1
+                    if bill.amount_untaxed > 0:
+                        tax_pcts.append(round(bill.amount_tax / bill.amount_untaxed * 100, 2))
+                else:
+                    tax_stats['bills_no_tax'] += 1
+            if tax_pcts:
+                tax_stats['avg_tax_pct'] = round(sum(tax_pcts) / len(tax_pcts), 2)
+            if tax_stats['total_bills'] > 0:
+                if tax_stats['bills_no_tax'] == tax_stats['total_bills']:
+                    tax_stats['tax_note'] = 'This vendor NEVER charges HST. All bills are zero-rated.'
+                elif tax_stats['avg_tax_pct'] < 2.0 and tax_stats['bills_with_tax'] > 0:
+                    tax_stats['tax_note'] = (
+                        f'HST only on shipping (avg {tax_stats["avg_tax_pct"]}%). '
+                        f'Do NOT apply HST to full amount.'
+                    )
+                elif tax_stats['avg_tax_pct'] >= 12.0:
+                    tax_stats['tax_note'] = f'Consistently charges HST at ~{tax_stats["avg_tax_pct"]}%.'
+            result['vendor_tax_pattern'] = tax_stats
+
+    return result
+
+
+def create_vendor_bill(env, params):
+    """[Tier 3] Create a vendor bill (account.move with move_type='in_invoice').
+    Requires user approval before execution."""
+    partner_id = int(params['partner_id'])
+    invoice_date = params.get('invoice_date', str(fields.Date.today()))
+    bill_lines = params.get('lines', [])
+    if not bill_lines:
+        return {'error': 'At least one invoice line is required'}
+
+    partner = env['res.partner'].browse(partner_id)
+    if not partner.exists():
+        return {'error': f'Partner not found: {partner_id}'}
+
+    invoice_line_vals = []
+    for line in bill_lines:
+        line_vals = {
+            'name': line.get('description', 'Expense'),
+            'price_unit': float(line.get('price_unit', 0)),
+            'quantity': float(line.get('quantity', 1)),
+        }
+        if line.get('account_id'):
+            line_vals['account_id'] = int(line['account_id'])
+        if line.get('tax_ids'):
+            line_vals['tax_ids'] = [(6, 0, [int(t) for t in line['tax_ids']])]
+        invoice_line_vals.append((0, 0, line_vals))
+
+    try:
+        bill = env['account.move'].create({
+            'move_type': 'in_invoice',
+            'partner_id': partner_id,
+            'invoice_date': invoice_date,
+            'date': invoice_date,
+            'invoice_line_ids': invoice_line_vals,
+            'company_id': env.company.id,
+        })
+
+        if params.get('post', False):
+            bill.action_post()
+
+        return {
+            'status': 'created',
+            'bill_id': bill.id,
+            'bill_name': bill.name,
+            'partner': partner.name,
+            'amount_total': bill.amount_total,
+            'state': bill.state,
+        }
+    except Exception as e:
+        _logger.error("Failed to create vendor bill: %s", e)
+        return {'error': str(e)}
+
+
+def register_bill_payment(env, params):
+    """[Tier 3] Register payment on a posted vendor bill and optionally reconcile to bank line.
+    Requires user approval before execution."""
+    bill_id = int(params['bill_id'])
+    journal_id = int(params['journal_id'])
+    bill = env['account.move'].browse(bill_id)
+    if not bill.exists() or bill.state != 'posted':
+        return {'error': 'Bill not found or not posted'}
+
+    payment_date = params.get('payment_date', str(fields.Date.today()))
+
+    try:
+        # Use the payment register wizard
+        ctx = {
+            'active_model': 'account.move',
+            'active_ids': [bill_id],
+        }
+        wizard = env['account.payment.register'].with_context(**ctx).create({
+            'journal_id': journal_id,
+            'payment_date': payment_date,
+        })
+        # Optionally set amount if provided (otherwise defaults to bill amount)
+        if params.get('amount'):
+            wizard.amount = float(params['amount'])
+
+        payments = wizard.action_create_payments()
+
+        # Find the created payment
+        payment = None
+        if isinstance(payments, dict) and payments.get('res_id'):
+            payment = env['account.payment'].browse(payments['res_id'])
+        elif isinstance(payments, dict) and payments.get('domain'):
+            payment = env['account.payment'].search(payments['domain'], limit=1)
+        else:
+            # Fallback: find the latest payment for this bill
+            payment = env['account.payment'].search([
+                ('partner_id', '=', bill.partner_id.id),
+            ], order='create_date desc', limit=1)
+
+        result = {
+            'status': 'paid',
+            'bill_id': bill_id,
+            'bill_name': bill.name,
+            'payment_state': bill.payment_state,
+        }
+        if payment:
+            result['payment_id'] = payment.id
+            result['payment_name'] = payment.name
+
+        # Optionally reconcile to a bank statement line
+        if params.get('statement_line_id') and payment:
+            try:
+                st_line = env['account.bank.statement.line'].browse(int(params['statement_line_id']))
+                if st_line.exists() and not st_line.is_reconciled:
+                    # Find the payment's move lines on the bank's outstanding account
+                    pay_move_lines = payment.move_id.line_ids.filtered(
+                        lambda l: l.account_id.reconcile and not l.reconciled
+                    )
+                    if pay_move_lines:
+                        st_line.set_line_bank_statement_line(pay_move_lines.ids)
+                        result['reconciled'] = True
+                        result['statement_line_id'] = st_line.id
+            except Exception as e:
+                _logger.warning("Payment created but bank reconciliation failed: %s", e)
+                result['reconcile_error'] = str(e)
+
+        return result
+    except Exception as e:
+        _logger.error("Failed to register payment: %s", e)
+        return {'error': str(e)}
+
+
+TOOLS = {
+    'get_ap_aging': get_ap_aging,
+    'find_duplicate_bills': find_duplicate_bills,
+    'match_bill_to_po': match_bill_to_po,
+    'get_unpaid_bills': get_unpaid_bills,
+    'verify_bill_taxes': verify_bill_taxes,
+    'get_payment_schedule': get_payment_schedule,
+    'search_partners': search_partners,
+    'find_similar_bank_lines': find_similar_bank_lines,
+    'create_vendor_bill': create_vendor_bill,
+    'register_bill_payment': register_bill_payment,
+}

fusion_accounting_ai/services/tools/accounts_receivable.py

@@ -0,0 +1,194 @@
+import logging
+from odoo import fields
+
+_logger = logging.getLogger(__name__)
+
+
+def get_ar_aging(env, params):
+    today = fields.Date.today()
+    domain = [
+        ('account_id.account_type', '=', 'asset_receivable'),
+        ('parent_state', '=', 'posted'),
+        ('reconciled', '=', False),
+        ('company_id', '=', env.company.id),
+    ]
+    amls = env['account.move.line'].search(domain)
+
+    buckets = {'current': 0, '1_30': 0, '31_60': 0, '61_90': 0, '90_plus': 0}
+    for aml in amls:
+        if not aml.date_maturity or aml.date_maturity >= today:
+            buckets['current'] += aml.amount_residual
+        else:
+            days = (today - aml.date_maturity).days
+            if days <= 30:
+                buckets['1_30'] += aml.amount_residual
+            elif days <= 60:
+                buckets['31_60'] += aml.amount_residual
+            elif days <= 90:
+                buckets['61_90'] += aml.amount_residual
+            else:
+                buckets['90_plus'] += aml.amount_residual
+
+    return {
+        'total': sum(buckets.values()),
+        'buckets': buckets,
+        'line_count': len(amls),
+    }
+
+
+def get_overdue_invoices(env, params):
+    today = fields.Date.today()
+    days_overdue = int(params.get('min_days_overdue', 1))
+    from datetime import timedelta
+    cutoff = today - timedelta(days=days_overdue)
+    invoices = env['account.move'].search([
+        ('move_type', '=', 'out_invoice'),
+        ('state', '=', 'posted'),
+        ('payment_state', 'in', ('not_paid', 'partial')),
+        ('invoice_date_due', '<', cutoff),
+        ('company_id', '=', env.company.id),
+    ], order='invoice_date_due asc', limit=int(params.get('limit', 50)))
+    return {
+        'count': len(invoices),
+        'invoices': [{
+            'id': inv.id,
+            'name': inv.name,
+            'partner': inv.partner_id.name if inv.partner_id else '',
+            'email': inv.partner_id.email or '' if inv.partner_id else '',
+            'phone': inv.partner_id.phone or '' if inv.partner_id else '',
+            'amount_total': inv.amount_total,
+            'amount_residual': inv.amount_residual,
+            'date_due': str(inv.invoice_date_due),
+            'days_overdue': (today - inv.invoice_date_due).days,
+        } for inv in invoices],
+    }
+
+
+def get_partner_balance(env, params):
+    """Get AR and AP balance for a partner. Accepts partner_id or partner_name."""
+    partner = None
+    if params.get('partner_id'):
+        partner = env['res.partner'].browse(int(params['partner_id']))
+    elif params.get('partner_name'):
+        partner = env['res.partner'].search([
+            ('name', 'ilike', params['partner_name']),
+        ], limit=1)
+    if not partner or not partner.exists():
+        return {'error': f"Partner not found: {params.get('partner_name', params.get('partner_id', '?'))}"}
+
+    # AR balance (receivable)
+    ar_amls = env['account.move.line'].search([
+        ('partner_id', '=', partner.id),
+        ('account_id.account_type', '=', 'asset_receivable'),
+        ('parent_state', '=', 'posted'),
+        ('reconciled', '=', False),
+        ('company_id', '=', env.company.id),
+    ])
+    ar_balance = sum(aml.amount_residual for aml in ar_amls)
+
+    # AP balance (payable)
+    ap_amls = env['account.move.line'].search([
+        ('partner_id', '=', partner.id),
+        ('account_id.account_type', '=', 'liability_payable'),
+        ('parent_state', '=', 'posted'),
+        ('reconciled', '=', False),
+        ('company_id', '=', env.company.id),
+    ])
+    ap_balance = sum(aml.amount_residual for aml in ap_amls)
+
+    open_items = [{
+        'id': aml.id,
+        'move_name': aml.move_id.name,
+        'ref': aml.ref or '',
+        'date': str(aml.date),
+        'amount_residual': aml.amount_residual,
+        'type': 'receivable' if aml.account_id.account_type == 'asset_receivable' else 'payable',
+        'date_maturity': str(aml.date_maturity) if aml.date_maturity else '',
+    } for aml in (ar_amls | ap_amls)[:30]]
+
+    return {
+        'partner': partner.name,
+        'partner_id': partner.id,
+        'ar_balance': ar_balance,
+        'ap_balance': ap_balance,
+        'net_balance': ar_balance + ap_balance,
+        'they_owe_us': ar_balance if ar_balance > 0 else 0,
+        'we_owe_them': abs(ap_balance) if ap_balance < 0 else 0,
+        'open_items': open_items,
+    }
+
+
+def send_followup(env, params):
+    partner_id = int(params['partner_id'])
+    partner = env['res.partner'].browse(partner_id)
+    if not partner.exists():
+        return {'error': 'Partner not found'}
+    options = {
+        'partner_id': partner_id,
+        'email': params.get('send_email', False),
+        'print': params.get('print_letter', False),
+        'sms': False,
+    }
+    if params.get('email_subject'):
+        options['email_subject'] = params['email_subject']
+    if params.get('body'):
+        options['body'] = params['body']
+    result = partner.execute_followup(options)
+    return {'status': 'sent', 'partner': partner.name, 'result': str(result) if result else 'done'}
+
+
+def get_followup_report(env, params):
+    partner_id = int(params['partner_id'])
+    partner = env['res.partner'].browse(partner_id)
+    if not partner.exists():
+        return {'error': 'Partner not found'}
+    try:
+        report = env['account.followup.report']
+        html = report._get_followup_report_html(partner)
+        return {'partner': partner.name, 'html': html}
+    except Exception as e:
+        return {'error': str(e)}
+
+
+def reconcile_payment_to_invoice(env, params):
+    move_line_ids = [int(x) for x in params['move_line_ids']]
+    amls = env['account.move.line'].browse(move_line_ids)
+    if len(amls) < 2:
+        return {'error': 'Need at least 2 journal items to reconcile'}
+    amls.reconcile()
+    return {
+        'status': 'reconciled',
+        'move_line_ids': move_line_ids,
+    }
+
+
+def get_unmatched_payments(env, params):
+    domain = [
+        ('account_id.account_type', '=', 'asset_receivable'),
+        ('parent_state', '=', 'posted'),
+        ('reconciled', '=', False),
+        ('move_id.payment_id', '!=', False),
+        ('company_id', '=', env.company.id),
+    ]
+    amls = env['account.move.line'].search(domain, order='date desc')
+    return {
+        'count': len(amls),
+        'payments': [{
+            'id': aml.id,
+            'date': str(aml.date),
+            'ref': aml.ref or aml.move_id.name,
+            'partner': aml.partner_id.name if aml.partner_id else '',
+            'amount': abs(aml.amount_residual),
+        } for aml in amls[:50]],
+    }
+
+
+TOOLS = {
+    'get_ar_aging': get_ar_aging,
+    'get_overdue_invoices': get_overdue_invoices,
+    'get_partner_balance': get_partner_balance,
+    'send_followup': send_followup,
+    'get_followup_report': get_followup_report,
+    'reconcile_payment_to_invoice': reconcile_payment_to_invoice,
+    'get_unmatched_payments': get_unmatched_payments,
+}

fusion_accounting_ai/services/tools/adp.py

@@ -0,0 +1,237 @@
+import logging
+from odoo import fields
+
+_logger = logging.getLogger(__name__)
+
+
+def get_adp_receivable_aging(env, params):
+    accounts = env['account.account'].search([
+        ('code', '=like', '1101%'),
+        ('company_ids', 'in', env.company.id),
+    ])
+    today = fields.Date.today()
+    amls = env['account.move.line'].search([
+        ('account_id', 'in', accounts.ids),
+        ('reconciled', '=', False),
+        ('parent_state', '=', 'posted'),
+    ])
+    buckets = {'current': 0, '1_30': 0, '31_60': 0, '61_90': 0, '90_plus': 0}
+    for aml in amls:
+        amt = abs(aml.amount_residual)
+        if not aml.date_maturity or aml.date_maturity >= today:
+            buckets['current'] += amt
+        else:
+            days = (today - aml.date_maturity).days
+            if days <= 30:
+                buckets['1_30'] += amt
+            elif days <= 60:
+                buckets['31_60'] += amt
+            elif days <= 90:
+                buckets['61_90'] += amt
+            else:
+                buckets['90_plus'] += amt
+    return {'total': sum(buckets.values()), 'buckets': buckets}
+
+
+def match_adp_payment_to_invoice(env, params):
+    move_line_ids = [int(x) for x in params['move_line_ids']]
+    amls = env['account.move.line'].browse(move_line_ids).exists()
+    if len(amls) < 2:
+        return {'error': 'Need at least 2 existing journal items to reconcile'}
+    amls.reconcile()
+    return {'status': 'matched', 'move_line_ids': amls.ids}
+
+
+def verify_adp_split(env, params):
+    invoice_id = int(params['invoice_id'])
+    invoice = env['account.move'].browse(invoice_id)
+    if not invoice.exists():
+        return {'error': 'Invoice not found'}
+    lines = invoice.invoice_line_ids
+    total = invoice.amount_untaxed
+    return {
+        'invoice': invoice.name,
+        'total_untaxed': total,
+        'total_with_tax': invoice.amount_total,
+        'lines': [{'name': l.name, 'subtotal': l.price_subtotal, 'total': l.price_total} for l in lines],
+        'balanced': abs(sum(l.price_subtotal for l in lines) - total) < 0.01,
+    }
+
+
+def find_adp_without_payment(env, params):
+    adp_partner = env['res.partner'].search([('name', 'ilike', 'ADP')], limit=1)
+    if not adp_partner:
+        return {'status': 'info', 'message': 'No ADP partner found in the system.'}
+    invoices = env['account.move'].search([
+        ('partner_id', '=', adp_partner.id),
+        ('move_type', '=', 'out_invoice'),
+        ('state', '=', 'posted'),
+        ('payment_state', 'in', ('not_paid', 'partial')),
+    ])
+    return {
+        'count': len(invoices),
+        'invoices': [{
+            'id': inv.id, 'name': inv.name,
+            'amount': inv.amount_residual, 'date': str(inv.date),
+        } for inv in invoices[:20]],
+    }
+
+
+def get_adp_summary(env, params):
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+    accounts = env['account.account'].search([
+        ('code', '=like', '1101%'), ('company_ids', 'in', env.company.id),
+    ])
+    domain = [
+        ('account_id', 'in', accounts.ids),
+        ('parent_state', '=', 'posted'),
+    ]
+    if date_from:
+        domain.append(('date', '>=', date_from))
+    if date_to:
+        domain.append(('date', '<=', date_to))
+    lines = env['account.move.line'].search(domain)
+    total_debit = sum(l.debit for l in lines)
+    total_credit = sum(l.credit for l in lines)
+    return {
+        'period': f'{date_from or "all"} to {date_to or "now"}',
+        'billed': total_debit,
+        'collected': total_credit,
+        'outstanding': total_debit - total_credit,
+    }
+
+
+def register_adp_batch_payment(env, params):
+    """Register payments for a batch of ADP invoices from a remittance advice.
+
+    Takes a list of invoice numbers with payment amounts and a payment date.
+    Registers a payment for each invoice via Odoo's payment wizard, which
+    creates outstanding receipt entries (PBNK2) on account 1050.
+
+    After calling this, use suggest_bank_line_matches on the bank deposit line
+    to match the outstanding receipts against the bank line.
+    """
+    invoices_data = params.get('invoices', [])
+    payment_date = params.get('payment_date')
+    journal_id = int(params.get('journal_id', 50))  # Default Scotia Current
+
+    if not invoices_data:
+        return {'error': 'No invoices provided'}
+    if not payment_date:
+        return {'error': 'payment_date is required (YYYY-MM-DD)'}
+
+    ADP_PARTNER_ID = 3421  # ADP (Assistive Device Program)
+    results = []
+    total_paid = 0.0
+    errors = []
+
+    for inv_data in invoices_data:
+        inv_number = str(inv_data.get('invoice_number', '')).strip()
+        amount = float(inv_data.get('amount', 0))
+
+        if not inv_number or not amount:
+            errors.append(f"Skipped: missing invoice_number or amount in {inv_data}")
+            continue
+
+        # Find the invoice by name/number
+        invoice = env['account.move'].search([
+            ('name', 'ilike', inv_number),
+            ('move_type', '=', 'out_invoice'),
+            ('state', '=', 'posted'),
+            ('company_id', '=', env.company.id),
+        ], limit=1)
+
+        if not invoice:
+            # Try without leading zeros or with different format
+            invoice = env['account.move'].search([
+                ('name', '=like', f'%{inv_number}'),
+                ('move_type', '=', 'out_invoice'),
+                ('state', '=', 'posted'),
+            ], limit=1)
+
+        if not invoice:
+            errors.append(f"Invoice {inv_number} not found")
+            continue
+
+        if invoice.payment_state == 'paid':
+            results.append({
+                'invoice': inv_number,
+                'status': 'already_paid',
+                'move_id': invoice.id,
+            })
+            continue
+
+        # Check if amount matches residual (allow partial)
+        if amount > invoice.amount_residual + 0.01:
+            errors.append(
+                f"Invoice {inv_number}: payment ${amount:.2f} exceeds "
+                f"residual ${invoice.amount_residual:.2f}"
+            )
+            continue
+
+        # Register payment via the payment wizard
+        try:
+            payment_vals = {
+                'payment_type': 'inbound',
+                'partner_type': 'customer',
+                'partner_id': invoice.partner_id.id or ADP_PARTNER_ID,
+                'amount': amount,
+                'date': payment_date,
+                'journal_id': journal_id,
+                'ref': f'ADP Remittance - {inv_number}',
+            }
+            # Use the payment register wizard
+            ctx = {
+                'active_model': 'account.move',
+                'active_ids': [invoice.id],
+            }
+            wizard = env['account.payment.register'].with_context(**ctx).create({
+                'payment_date': payment_date,
+                'amount': amount,
+                'journal_id': journal_id,
+                'payment_method_line_id': env['account.payment.method.line'].search([
+                    ('journal_id', '=', journal_id),
+                    ('payment_type', '=', 'inbound'),
+                ], limit=1).id,
+            })
+            wizard.action_create_payments()
+
+            results.append({
+                'invoice': inv_number,
+                'status': 'paid',
+                'amount': amount,
+                'move_id': invoice.id,
+                'move_name': invoice.name,
+            })
+            total_paid += amount
+        except Exception as e:
+            _logger.warning("ADP payment failed for %s: %s", inv_number, e)
+            errors.append(f"Invoice {inv_number}: payment failed — {e}")
+
+    env.cr.commit()
+
+    return {
+        'status': 'completed',
+        'paid_count': len([r for r in results if r.get('status') == 'paid']),
+        'already_paid_count': len([r for r in results if r.get('status') == 'already_paid']),
+        'total_paid': total_paid,
+        'results': results,
+        'errors': errors,
+        'message': (
+            f"Registered payments for {len([r for r in results if r.get('status') == 'paid'])} invoices "
+            f"totalling ${total_paid:,.2f}. "
+            + (f"{len(errors)} errors." if errors else "No errors.")
+            + " Now use suggest_bank_line_matches to match the bank deposit."
+        ),
+    }
+
+
+TOOLS = {
+    'get_adp_receivable_aging': get_adp_receivable_aging,
+    'match_adp_payment_to_invoice': match_adp_payment_to_invoice,
+    'verify_adp_split': verify_adp_split,
+    'find_adp_without_payment': find_adp_without_payment,
+    'get_adp_summary': get_adp_summary,
+    'register_adp_batch_payment': register_adp_batch_payment,
+}

fusion_accounting_ai/services/tools/audit.py

@@ -0,0 +1,164 @@
+import logging
+from odoo import fields
+
+_logger = logging.getLogger(__name__)
+
+
+def audit_posted_entry(env, params):
+    move_id = int(params['move_id'])
+    move = env['account.move'].browse(move_id)
+    if not move.exists():
+        return {'error': 'Entry not found'}
+    issues = []
+    total_debit = sum(l.debit for l in move.line_ids)
+    total_credit = sum(l.credit for l in move.line_ids)
+    if abs(total_debit - total_credit) > 0.01:
+        issues.append({'severity': 'critical', 'issue': f'Unbalanced entry: debit={total_debit}, credit={total_credit}'})
+    for line in move.line_ids:
+        if not line.account_id:
+            issues.append({'severity': 'critical', 'issue': f'Line missing account: {line.name}'})
+    if not move.line_ids:
+        issues.append({'severity': 'warning', 'issue': 'Entry has no lines'})
+    return {
+        'move': move.name, 'date': str(move.date),
+        'issues': issues, 'clean': len(issues) == 0,
+    }
+
+
+def audit_account_balances(env, params):
+    from .journal_review import find_wrong_direction_balances
+    return find_wrong_direction_balances(env, params)
+
+
+def audit_tax_compliance(env, params):
+    from .hst_management import find_missing_tax_invoices, find_missing_itc_bills
+    invoices = find_missing_tax_invoices(env, params)
+    bills = find_missing_itc_bills(env, params)
+    return {
+        'missing_tax_invoices': invoices.get('missing_tax_count', 0),
+        'missing_itc_bills': bills.get('missing_itc_count', 0),
+        'total_issues': invoices.get('missing_tax_count', 0) + bills.get('missing_itc_count', 0),
+    }
+
+
+def audit_reconciliation_integrity(env, params):
+    from .journal_review import verify_reconciliation_integrity
+    return verify_reconciliation_integrity(env, params)
+
+
+def check_hash_chain(env, params):
+    from .month_end import run_hash_integrity_check
+    return run_hash_integrity_check(env, params)
+
+
+def check_sequence_gaps(env, params):
+    from .journal_review import find_sequence_gaps
+    return find_sequence_gaps(env, params)
+
+
+def flag_entry(env, params):
+    move_id = int(params['move_id'])
+    flag = params.get('flag', 'Review Required')
+    recommendation = params.get('recommendation', '')
+    move = env['account.move'].browse(move_id)
+    if not move.exists():
+        return {'error': 'Entry not found'}
+    body = f'<strong>🏴 {flag}</strong><br/>{recommendation}'
+    move.message_post(body=body, message_type='comment', subtype_xmlid='mail.mt_note')
+    return {'status': 'flagged', 'move': move.name, 'flag': flag}
+
+
+def get_audit_status(env, params):
+    try:
+        AuditStatus = env['account.audit.account.status']
+    except KeyError:
+        return {'error': 'Audit status model (account.audit.account.status) is not available. The account_audit Enterprise module may not be installed.'}
+    statuses = AuditStatus.search([])
+    return {
+        'statuses': [{
+            'id': s.id,
+            'account': s.account_id.name,
+            'status': s.status,
+            'audit': s.audit_id.display_name if s.audit_id else '',
+        } for s in statuses[:50]],
+    }
+
+
+def set_audit_status(env, params):
+    try:
+        AuditStatus = env['account.audit.account.status']
+    except KeyError:
+        return {'error': 'Audit status model (account.audit.account.status) is not available. The account_audit Enterprise module may not be installed.'}
+    status_id = int(params['status_id'])
+    new_status = params['status']
+    rec = AuditStatus.browse(status_id)
+    if not rec.exists():
+        return {'error': 'Audit status record not found'}
+    rec.status = new_status
+    return {'status': 'updated', 'id': status_id, 'new_status': new_status}
+
+
+def get_audit_trail(env, params):
+    move_id = int(params['move_id'])
+    move = env['account.move'].browse(move_id)
+    if not move.exists():
+        return {'error': 'Entry not found'}
+    messages = env['mail.message'].search([
+        ('model', '=', 'account.move'),
+        ('res_id', '=', move_id),
+    ], order='date desc', limit=20)
+    return {
+        'move': move.name,
+        'messages': [{
+            'date': str(m.date),
+            'author': m.author_id.name if m.author_id else '',
+            'body': m.body or '',
+            'type': m.message_type,
+        } for m in messages],
+    }
+
+
+def run_full_audit(env, params):
+    results = {}
+    results['account_balances'] = audit_account_balances(env, params)
+    results['tax_compliance'] = audit_tax_compliance(env, params)
+    results['reconciliation'] = audit_reconciliation_integrity(env, params)
+    results['hash_chain'] = check_hash_chain(env, params)
+    results['sequence_gaps'] = check_sequence_gaps(env, params)
+
+    total_issues = 0
+    for key, val in results.items():
+        total_issues += val.get('count', 0) + val.get('total_issues', 0)
+
+    score = max(0, 100 - total_issues * 5)
+    return {
+        'score': min(100, score),
+        'total_issues': total_issues,
+        'details': results,
+    }
+
+
+def get_audit_report(env, params):
+    audit = run_full_audit(env, params)
+    report_lines = [f"Audit Score: {audit['score']}/100", f"Total Issues: {audit['total_issues']}", '']
+    for domain, detail in audit.get('details', {}).items():
+        report_lines.append(f"--- {domain.replace('_', ' ').title()} ---")
+        count = detail.get('count', detail.get('total_issues', 0))
+        report_lines.append(f"  Issues: {count}")
+    return {'report': '\n'.join(report_lines), 'score': audit['score']}
+
+
+TOOLS = {
+    'audit_posted_entry': audit_posted_entry,
+    'audit_account_balances': audit_account_balances,
+    'audit_tax_compliance': audit_tax_compliance,
+    'audit_reconciliation_integrity': audit_reconciliation_integrity,
+    'check_hash_chain': check_hash_chain,
+    'check_sequence_gaps': check_sequence_gaps,
+    'flag_entry': flag_entry,
+    'get_audit_status': get_audit_status,
+    'set_audit_status': set_audit_status,
+    'get_audit_trail': get_audit_trail,
+    'run_full_audit': run_full_audit,
+    'get_audit_report': get_audit_report,
+}

fusion_accounting_ai/services/tools/bank_reconciliation.py

@@ -0,0 +1,961 @@
+import logging
+from datetime import datetime
+from odoo import fields
+
+_logger = logging.getLogger(__name__)
+
+
+def get_unreconciled_bank_lines(env, params):
+    domain = [('is_reconciled', '=', False), ('company_id', '=', env.company.id)]
+    if params.get('journal_id'):
+        domain.append(('journal_id', '=', int(params['journal_id'])))
+    if params.get('date_from'):
+        domain.append(('date', '>=', params['date_from']))
+    if params.get('date_to'):
+        domain.append(('date', '<=', params['date_to']))
+    if params.get('min_amount'):
+        domain.append(('amount', '>=', float(params['min_amount'])))
+    limit = int(params.get('limit', 50))
+    lines = env['account.bank.statement.line'].search(domain, limit=limit, order='date desc')
+    return {
+        'count': len(lines),
+        'total_amount': sum(abs(l.amount) for l in lines),
+        'lines': [{
+            'id': l.id,
+            'date': str(l.date),
+            'payment_ref': l.payment_ref or '',
+            'partner_name': l.partner_name or (l.partner_id.name if l.partner_id else ''),
+            'amount': l.amount,
+            'journal': l.journal_id.name,
+        } for l in lines],
+    }
+
+
+def get_unreconciled_receipts(env, params):
+    account_code = params.get('account_code', '1122')
+    accounts = env['account.account'].search([
+        ('code', '=like', f'{account_code}%'),
+        ('company_ids', 'in', env.company.id),
+    ])
+    domain = [
+        ('account_id', 'in', accounts.ids),
+        ('reconciled', '=', False),
+        ('parent_state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ]
+    lines = env['account.move.line'].search(domain, order='date desc')
+    return {
+        'count': len(lines),
+        'total_amount': sum(abs(l.amount_residual) for l in lines),
+        'lines': [{
+            'id': l.id,
+            'date': str(l.date),
+            'ref': l.ref or l.move_id.name,
+            'partner': l.partner_id.name if l.partner_id else '',
+            'amount_residual': l.amount_residual,
+        } for l in lines],
+    }
+
+
+def match_bank_line_to_payments(env, params):
+    st_line_id = int(params['statement_line_id'])
+    move_line_ids = [int(x) for x in params['move_line_ids']]
+    st_line = env['account.bank.statement.line'].browse(st_line_id)
+    if not st_line.exists():
+        return {'error': 'Statement line not found'}
+    st_line.set_line_bank_statement_line(move_line_ids)
+    return {
+        'status': 'matched',
+        'statement_line_id': st_line_id,
+        'matched_move_lines': move_line_ids,
+        'is_reconciled': st_line.is_reconciled,
+    }
+
+
+def auto_reconcile_bank_lines(env, params):
+    company_id = params.get('company_id', env.company.id)
+    lines = env['account.bank.statement.line'].search([
+        ('is_reconciled', '=', False),
+        ('company_id', '=', int(company_id)),
+    ])
+    before_count = len(lines)
+    lines._try_auto_reconcile_statement_lines(company_id=int(company_id))
+    still_unreconciled = env['account.bank.statement.line'].search([
+        ('is_reconciled', '=', False),
+        ('company_id', '=', int(company_id)),
+    ])
+    reconciled_count = before_count - len(still_unreconciled)
+    return {
+        'status': 'completed',
+        'lines_before': before_count,
+        'lines_reconciled': reconciled_count,
+        'lines_remaining': len(still_unreconciled),
+    }
+
+
+def apply_reconcile_model(env, params):
+    model_id = int(params['model_id'])
+    st_line_id = int(params['statement_line_id'])
+    reco_model = env['account.reconcile.model'].browse(model_id)
+    st_line = env['account.bank.statement.line'].browse(st_line_id)
+    if not reco_model.exists() or not st_line.exists():
+        return {'error': 'Model or statement line not found'}
+    _liquidity_lines, suspense_lines, _other_lines = st_line._seek_for_lines()
+    residual = sum(l.amount_residual for l in suspense_lines) if suspense_lines else st_line.amount
+    write_off_vals = reco_model._get_write_off_move_lines_dict(st_line, residual)
+    if write_off_vals:
+        line_ids_create_command = [(0, 0, vals) for vals in write_off_vals]
+        st_line.move_id.write({'line_ids': line_ids_create_command})
+    return {
+        'status': 'applied',
+        'model': reco_model.name,
+        'write_off_lines': len(write_off_vals) if write_off_vals else 0,
+    }
+
+
+def unmatch_bank_line(env, params):
+    st_line_id = int(params['statement_line_id'])
+    st_line = env['account.bank.statement.line'].browse(st_line_id)
+    if not st_line.exists():
+        return {'error': 'Statement line not found'}
+    st_line.action_unreconcile_entry()
+    return {'status': 'unmatched', 'statement_line_id': st_line_id}
+
+
+def get_reconcile_suggestions(env, params):
+    st_line_id = int(params['statement_line_id'])
+    st_line = env['account.bank.statement.line'].browse(st_line_id)
+    if not st_line.exists():
+        return {'error': 'Statement line not found'}
+    models = env['account.reconcile.model'].search([
+        ('company_id', '=', env.company.id),
+    ])
+    return {
+        'models': [{
+            'id': m.id,
+            'name': m.name,
+            'trigger': m.trigger if hasattr(m, 'trigger') else 'manual',
+        } for m in models],
+    }
+
+
+def sum_payments_by_date(env, params):
+    """Sum payment/journal activity for a date range.
+    IMPORTANT: Always pass journal_ids to filter to specific journals.
+    Without journal_ids, returns totals across ALL journals which is
+    almost never what you want for reconciliation."""
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+    if not date_from or not date_to:
+        return {'error': 'date_from and date_to are required'}
+    journal_ids = params.get('journal_ids', [])
+    domain = [
+        ('parent_state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+        ('date', '>=', date_from),
+        ('date', '<=', date_to),
+    ]
+    scope = 'all journals'
+    if journal_ids:
+        jids = [int(j) for j in journal_ids]
+        domain.append(('journal_id', 'in', jids))
+        journals = env['account.journal'].browse(jids)
+        scope = ', '.join(j.name for j in journals if j.exists())
+    else:
+        # Without journal filter, include a warning and break down by journal
+        pass
+
+    lines = env['account.move.line'].search(domain)
+    total_debit = sum(l.debit for l in lines)
+    total_credit = sum(l.credit for l in lines)
+
+    result = {
+        'date_from': date_from,
+        'date_to': date_to,
+        'total_debit': total_debit,
+        'total_credit': total_credit,
+        'net': total_debit - total_credit,
+        'line_count': len(lines),
+        'scope': scope,
+    }
+
+    # If no journal filter, add per-journal breakdown so AI doesn't
+    # mistake company-wide totals for a specific journal's activity
+    if not journal_ids:
+        result['warning'] = (
+            'No journal_ids filter was provided. These totals are across ALL '
+            'journals in the company. To get card payment totals, pass the '
+            'specific card/POS journal IDs.'
+        )
+        journal_totals = {}
+        for l in lines:
+            jname = l.journal_id.name
+            if jname not in journal_totals:
+                journal_totals[jname] = {'debit': 0.0, 'credit': 0.0, 'count': 0}
+            journal_totals[jname]['debit'] += l.debit
+            journal_totals[jname]['credit'] += l.credit
+            journal_totals[jname]['count'] += 1
+        result['by_journal'] = [
+            {'journal': jn, 'debit': v['debit'], 'credit': v['credit'], 'count': v['count']}
+            for jn, v in sorted(journal_totals.items(), key=lambda x: -x[1]['debit'])
+        ][:15]
+
+    return result
+
+
+def get_bank_line_details(env, params):
+    """Get full details of a single bank statement line plus matching suggestions."""
+    line_id = int(params['line_id'])
+    line = env['account.bank.statement.line'].browse(line_id)
+    if not line.exists():
+        return {'error': 'Bank statement line not found'}
+
+    result = {
+        'id': line.id,
+        'date': str(line.date),
+        'payment_ref': line.payment_ref or '',
+        'partner_name': line.partner_name or (line.partner_id.name if line.partner_id else ''),
+        'partner_id': line.partner_id.id if line.partner_id else None,
+        'amount': line.amount,
+        'journal': line.journal_id.name,
+        'journal_id': line.journal_id.id,
+        'is_reconciled': line.is_reconciled,
+        'existing_bills': [],
+        'suggested_partner': None,
+    }
+
+    # Search for existing vendor bills matching amount ± $0.50 and date ± 3 days
+    abs_amount = abs(line.amount)
+    from datetime import timedelta as td
+    date_from = line.date - td(days=3)
+    date_to = line.date + td(days=3)
+    matching_bills = env['account.move'].search([
+        ('move_type', '=', 'in_invoice'),
+        ('state', '=', 'posted'),
+        ('amount_total', '>=', abs_amount - 0.50),
+        ('amount_total', '<=', abs_amount + 0.50),
+        ('date', '>=', str(date_from)),
+        ('date', '<=', str(date_to)),
+        ('company_id', '=', env.company.id),
+    ], limit=5)
+    for bill in matching_bills:
+        result['existing_bills'].append({
+            'id': bill.id,
+            'name': bill.name,
+            'partner': bill.partner_id.name if bill.partner_id else '',
+            'amount_total': bill.amount_total,
+            'date': str(bill.date),
+            'payment_state': bill.payment_state,
+        })
+
+    # Try to suggest a partner from payment_ref keyword
+    if line.payment_ref and not line.partner_id:
+        # Extract meaningful words from payment_ref (skip common banking terms)
+        skip_words = {'misc', 'payment', 'online', 'banking', 'pad', 'business',
+                      'deposit', 'cheque', 'transfer', 'e-transfer', 'sent', 'autodeposit'}
+        words = [w for w in line.payment_ref.split() if len(w) > 2 and w.lower() not in skip_words]
+        for word in words[:3]:
+            partners = env['res.partner'].search([
+                ('name', 'ilike', word),
+                ('supplier_rank', '>', 0),
+            ], limit=3)
+            if partners:
+                result['suggested_partner'] = {
+                    'id': partners[0].id,
+                    'name': partners[0].name,
+                    'match_word': word,
+                }
+                break
+
+    return result
+
+
+def check_recurring_pattern(env, params):
+    """Check if a bank line matches a known recurring payment pattern.
+    Returns the historical coding (account, HST, partner, reconcile model) if found."""
+    line_id = params.get('line_id')
+    payment_ref = params.get('payment_ref', '')
+    amount = params.get('amount')
+
+    # If line_id provided, get the ref and amount from the line
+    if line_id:
+        line = env['account.bank.statement.line'].browse(int(line_id))
+        if line.exists():
+            payment_ref = line.payment_ref or ''
+            amount = line.amount
+
+    if not payment_ref:
+        return {'match': False, 'reason': 'No payment reference to match'}
+
+    # Search cached patterns by keyword
+    patterns = env['fusion.recurring.pattern'].search([
+        ('company_id', '=', env.company.id),
+    ])
+
+    best_match = None
+    for pat in patterns:
+        if not pat.ref_keyword:
+            continue
+        # Check if the pattern keyword appears in the payment_ref
+        if pat.ref_keyword.lower()[:30] in payment_ref.lower():
+            # If amount matches too, it's a strong match
+            if amount and pat.amount_is_fixed and abs(pat.amount - amount) < 0.01:
+                best_match = pat
+                break
+            # Keyword-only match (amount may vary)
+            if not best_match or pat.occurrences > best_match.occurrences:
+                best_match = pat
+
+    if not best_match:
+        return {'match': False, 'payment_ref': payment_ref}
+
+    result = {
+        'match': True,
+        'pattern_id': best_match.id,
+        'pattern_name': best_match.name,
+        'occurrences': best_match.occurrences,
+        'first_seen': str(best_match.first_seen) if best_match.first_seen else '',
+        'last_seen': str(best_match.last_seen) if best_match.last_seen else '',
+        'expense_account_id': best_match.expense_account_id.id if best_match.expense_account_id else None,
+        'expense_account_code': best_match.expense_account_code or '',
+        'expense_account_name': best_match.expense_account_id.name if best_match.expense_account_id else '',
+        'has_hst': best_match.has_hst,
+        'partner_id': best_match.partner_id.id if best_match.partner_id else None,
+        'partner_name': best_match.partner_id.name if best_match.partner_id else '',
+        'action_note': best_match.action_note or '',
+        'amount_is_fixed': best_match.amount_is_fixed,
+    }
+    if best_match.reconcile_model_id:
+        result['reconcile_model_id'] = best_match.reconcile_model_id.id
+        result['reconcile_model_name'] = best_match.reconcile_model_id.name
+
+    return result
+
+
+def match_internal_transfers(env, params):
+    """[Tier 3] Find and match inter-account transfers between two bank journals.
+    Matches exact amounts within a date window. Only matches when there is exactly
+    ONE candidate on each side (no ambiguous matches). Requires user approval.
+
+    Typical use: Scotia Current Account ↔ Scotia Visa payments."""
+    journal_a_id = int(params['journal_a_id'])  # e.g., Scotia Current (50)
+    journal_b_id = int(params['journal_b_id'])  # e.g., Scotia Visa (51)
+    date_from = params.get('date_from', '2025-01-01')
+    date_to = params.get('date_to', '2025-03-31')
+    max_days_apart = int(params.get('max_days_apart', 2))
+
+    # Get unreconciled positive lines from both journals
+    # (transfers show as positive on the RECEIVING side)
+    lines_a = env['account.bank.statement.line'].search([
+        ('is_reconciled', '=', False),
+        ('journal_id', '=', journal_a_id),
+        ('company_id', '=', env.company.id),
+    ])
+    lines_a = lines_a.filtered(
+        lambda l: l.move_id.date >= fields.Date.from_string(date_from)
+                  and l.move_id.date <= fields.Date.from_string(date_to)
+                  and l.amount > 0  # money coming IN on this account
+    )
+
+    lines_b = env['account.bank.statement.line'].search([
+        ('is_reconciled', '=', False),
+        ('journal_id', '=', journal_b_id),
+        ('company_id', '=', env.company.id),
+    ])
+    lines_b = lines_b.filtered(
+        lambda l: l.move_id.date >= fields.Date.from_string(date_from)
+                  and l.move_id.date <= fields.Date.from_string(date_to)
+                  and l.amount > 0  # money coming IN on this account
+    )
+
+    matched_pairs = []
+    used_a = set()
+    used_b = set()
+
+    # For each line in A, find exact-amount match in B within date window
+    for la in sorted(lines_a, key=lambda l: l.move_id.date):
+        if la.id in used_a:
+            continue
+        candidates = []
+        for lb in lines_b:
+            if lb.id in used_b:
+                continue
+            if abs(la.amount - lb.amount) < 0.01:
+                days = abs((la.move_id.date - lb.move_id.date).days)
+                if days <= max_days_apart:
+                    candidates.append(lb)
+        # Only match if EXACTLY ONE candidate — skip ambiguous
+        if len(candidates) == 1:
+            lb = candidates[0]
+            matched_pairs.append({
+                'line_a_id': la.id,
+                'line_a_date': str(la.move_id.date),
+                'line_a_ref': la.payment_ref or '',
+                'line_a_journal': la.journal_id.name,
+                'line_b_id': lb.id,
+                'line_b_date': str(lb.move_id.date),
+                'line_b_ref': lb.payment_ref or '',
+                'line_b_journal': lb.journal_id.name,
+                'amount': la.amount,
+                'days_apart': abs((la.move_id.date - lb.move_id.date).days),
+            })
+            used_a.add(la.id)
+            used_b.add(lb.id)
+
+    if not matched_pairs:
+        return {
+            'status': 'no_matches',
+            'message': 'No unambiguous transfer pairs found.',
+            'lines_a_checked': len(lines_a),
+            'lines_b_checked': len(lines_b),
+        }
+
+    # If this is just a dry-run check (no execute flag), return the pairs for review
+    if not params.get('execute', False):
+        return {
+            'status': 'pairs_found',
+            'count': len(matched_pairs),
+            'pairs': matched_pairs,
+            'message': f'Found {len(matched_pairs)} unambiguous transfer pairs. Set execute=true to reconcile them.',
+        }
+
+    # Execute: create internal transfer journal entries to reconcile both sides
+    reconciled = []
+    for pair in matched_pairs:
+        try:
+            line_a = env['account.bank.statement.line'].browse(pair['line_a_id'])
+            line_b = env['account.bank.statement.line'].browse(pair['line_b_id'])
+
+            # Create an internal transfer payment
+            payment = env['account.payment'].create({
+                'payment_type': 'outbound',
+                'partner_type': 'supplier',
+                'partner_id': env.company.partner_id.id,  # Self as partner for internal transfer
+                'amount': pair['amount'],
+                'journal_id': journal_a_id,
+                'destination_journal_id': journal_b_id,
+                'date': line_a.move_id.date,
+                'ref': f'Internal Transfer: {pair["line_a_ref"]} ↔ {pair["line_b_ref"]}',
+                'is_internal_transfer': True,
+            })
+            payment.action_post()
+
+            # Now match the payment's move lines to the bank statement lines
+            # The payment creates lines on both journals' outstanding accounts
+            for move_line in payment.move_id.line_ids:
+                if move_line.journal_id.id == journal_a_id and not move_line.reconciled:
+                    try:
+                        line_a.set_line_bank_statement_line(move_line.ids)
+                    except Exception:
+                        pass
+            # Check paired transfer for the other side
+            if payment.paired_internal_transfer_payment_id:
+                paired = payment.paired_internal_transfer_payment_id
+                for move_line in paired.move_id.line_ids:
+                    if move_line.journal_id.id == journal_b_id and not move_line.reconciled:
+                        try:
+                            line_b.set_line_bank_statement_line(move_line.ids)
+                        except Exception:
+                            pass
+
+            reconciled.append({
+                'line_a_id': pair['line_a_id'],
+                'line_b_id': pair['line_b_id'],
+                'amount': pair['amount'],
+                'payment_id': payment.id,
+                'status': 'reconciled',
+            })
+        except Exception as e:
+            _logger.error("Failed to reconcile transfer pair %s: %s", pair, e)
+            reconciled.append({
+                'line_a_id': pair['line_a_id'],
+                'line_b_id': pair['line_b_id'],
+                'amount': pair['amount'],
+                'status': 'error',
+                'error': str(e),
+            })
+
+    return {
+        'status': 'executed',
+        'total_pairs': len(matched_pairs),
+        'reconciled': len([r for r in reconciled if r['status'] == 'reconciled']),
+        'errors': len([r for r in reconciled if r['status'] == 'error']),
+        'details': reconciled,
+    }
+
+
+def find_unreconciled_cheques(env, params):
+    """Find unreconciled cheque bank lines and classify as payroll vs non-payroll
+    by checking if the amount matches an existing payroll liability entry."""
+    PAYROLL_ACCT = 433  # 2201 Payroll Liabilities
+    journal_id = int(params.get('journal_id', 50))  # Default Scotia Current
+    limit = int(params.get('limit', 50))
+
+    AML = env['account.move.line'].sudo()
+    BSL = env['account.bank.statement.line'].sudo()
+
+    # Build set of known payroll liability amounts
+    payroll_amounts = set()
+    for aml in AML.search([
+        ('account_id', '=', PAYROLL_ACCT),
+        ('parent_state', '=', 'posted'),
+        ('credit', '>', 0),
+    ]):
+        payroll_amounts.add(round(aml.credit, 2))
+
+    cheque_lines = BSL.search([
+        ('journal_id', '=', journal_id),
+        ('is_reconciled', '=', False),
+        ('payment_ref', 'ilike', 'cheque'),
+        ('amount', '<', 0),
+        ('company_id', '=', env.company.id),
+    ], limit=limit, order='move_id desc')
+
+    payroll = []
+    non_payroll = []
+    for line in cheque_lines:
+        amt = round(abs(line.amount), 2)
+        entry = {
+            'id': line.id,
+            'date': str(line.move_id.date),
+            'ref': line.payment_ref or '',
+            'amount': amt,
+            'journal': line.journal_id.name,
+        }
+        if amt in payroll_amounts:
+            entry['type'] = 'payroll'
+            payroll.append(entry)
+        else:
+            entry['type'] = 'non_payroll'
+            non_payroll.append(entry)
+
+    return {
+        'count': len(cheque_lines),
+        'payroll_count': len(payroll),
+        'non_payroll_count': len(non_payroll),
+        'payroll': payroll,
+        'non_payroll': non_payroll,
+    }
+
+
+def reconcile_payroll_cheques(env, params):
+    """Reconcile payroll cheque bank lines by applying the Payroll Cheque Clearing
+    reconcile model. Only reconciles cheques whose amount matches an existing
+    payroll liability entry on account 2201. Non-payroll cheques are skipped.
+
+    Params:
+        journal_id (int): Bank journal ID (default 50 = Scotia Current)
+        line_ids (list): Optional list of specific bank line IDs to reconcile.
+                         If not provided, reconciles all matching payroll cheques.
+    """
+    PAYROLL_ACCT = 433
+    journal_id = int(params.get('journal_id', 50))
+
+    AML = env['account.move.line'].sudo()
+    BSL = env['account.bank.statement.line'].sudo()
+    RecModel = env['account.reconcile.model'].sudo()
+
+    model = RecModel.search([
+        ('name', 'ilike', 'Payroll Cheque'),
+        ('company_id', '=', env.company.id),
+    ], limit=1)
+    if not model:
+        return {'error': 'No "Payroll Cheque Clearing" reconcile model found. Create one first.'}
+
+    # Get lines to process
+    if params.get('line_ids'):
+        cheque_lines = BSL.browse([int(x) for x in params['line_ids']])
+        cheque_lines = cheque_lines.filtered(lambda l: not l.is_reconciled)
+    else:
+        cheque_lines = BSL.search([
+            ('journal_id', '=', journal_id),
+            ('is_reconciled', '=', False),
+            ('payment_ref', 'ilike', 'cheque'),
+            ('amount', '<', 0),
+            ('company_id', '=', env.company.id),
+        ])
+
+    # Filter post-lock-date
+    lock = env.company.fiscalyear_lock_date
+    if lock:
+        cheque_lines = cheque_lines.filtered(lambda l: l.move_id.date > lock)
+
+    # Filter to payroll-only amounts
+    payroll_amounts = set()
+    for aml in AML.search([
+        ('account_id', '=', PAYROLL_ACCT),
+        ('parent_state', '=', 'posted'),
+        ('credit', '>', 0),
+    ]):
+        payroll_amounts.add(round(aml.credit, 2))
+
+    payroll_lines = cheque_lines.filtered(
+        lambda l: round(abs(l.amount), 2) in payroll_amounts
+    )
+    skipped = len(cheque_lines) - len(payroll_lines)
+
+    if not payroll_lines:
+        return {
+            'status': 'nothing_to_do',
+            'message': f'No payroll cheques to reconcile ({skipped} non-payroll cheques skipped)',
+        }
+
+    try:
+        model._apply_reconcile_models(payroll_lines)
+        env.cr.commit()
+    except Exception as e:
+        return {'error': f'Reconciliation failed: {e}'}
+
+    still = payroll_lines.filtered(lambda l: not l.is_reconciled)
+    reconciled = len(payroll_lines) - len(still)
+
+    return {
+        'status': 'completed',
+        'reconciled': reconciled,
+        'still_unreconciled': len(still),
+        'non_payroll_skipped': skipped,
+        'message': f'Reconciled {reconciled} payroll cheques. {skipped} non-payroll cheques skipped.',
+    }
+
+
+def _extract_partner_from_ref(env, payment_ref):
+    """Extract a partner from a bank line payment_ref using keyword matching."""
+    if not payment_ref:
+        return None
+    skip_words = {
+        'misc', 'payment', 'online', 'banking', 'pad', 'business', 'deposit',
+        'cheque', 'transfer', 'e-transfer', 'sent', 'autodeposit', 'credit',
+        'debit', 'memo', 'free', 'interac', 'from', 'the', 'and', 'for',
+        'miscellaneous', 'bill', 'correction', 'adjustment', 'other',
+    }
+    # Strip common suffixes like colons and split
+    clean_ref = payment_ref.replace(':', ' ').replace('-', ' ')
+    words = [w for w in clean_ref.split() if len(w) > 2 and w.lower() not in skip_words]
+    # Try progressively shorter phrases
+    for n in range(min(len(words), 4), 0, -1):
+        for i in range(len(words) - n + 1):
+            phrase = ' '.join(words[i:i+n])
+            partners = env['res.partner'].search([
+                ('name', 'ilike', phrase),
+                ('company_id', 'in', [env.company.id, False]),
+            ], limit=3)
+            if partners:
+                return partners[0]
+    # Fallback: try each word individually with supplier/customer rank
+    for word in words:
+        if len(word) < 4:
+            continue
+        partners = env['res.partner'].search([
+            ('name', 'ilike', word),
+            ('company_id', 'in', [env.company.id, False]),
+            '|', ('customer_rank', '>', 0), ('supplier_rank', '>', 0),
+        ], limit=3)
+        if partners:
+            return partners[0]
+    return None
+
+
+def _find_best_subset(candidates, target, max_items=8):
+    """Find the subset of candidates whose amounts sum closest to target.
+    Returns (aml_ids, total) for the best combination."""
+    items = candidates[:max_items]
+    if not items:
+        return [], 0.0
+    best_ids = []
+    best_total = 0.0
+    best_diff = abs(target)
+    n = len(items)
+    # Brute force all subsets (2^n, max 256)
+    for mask in range(1, 1 << n):
+        subset_ids = []
+        subset_total = 0.0
+        for j in range(n):
+            if mask & (1 << j):
+                subset_ids.append(items[j]['aml_id'])
+                subset_total += items[j]['amount_residual']
+        diff = abs(subset_total - target)
+        if diff < best_diff:
+            best_diff = diff
+            best_ids = subset_ids
+            best_total = subset_total
+        if diff < 0.01:
+            break  # Exact match found
+    return best_ids, round(best_total, 2)
+
+
+def suggest_bank_line_matches(env, params):
+    """Find candidate journal items (invoices/bills) that could match a bank statement line.
+    Scores and ranks matches, finds best subset-sum combination.
+    Returns data for a reconciliation-mode fusion-table."""
+    line_id = int(params['statement_line_id'])
+    line = env['account.bank.statement.line'].browse(line_id)
+    if not line.exists():
+        return {'error': 'Bank statement line not found'}
+    if line.is_reconciled:
+        return {'error': 'Bank statement line is already reconciled'}
+
+    AML = env['account.move.line'].sudo()
+    bank_amount = abs(line.amount)
+    line_date = line.move_id.date
+    is_incoming = line.amount > 0  # positive = customer payment, negative = vendor payment
+    from datetime import timedelta as td
+
+    # Determine partner
+    partner = line.partner_id
+    if not partner:
+        partner = _extract_partner_from_ref(env, line.payment_ref)
+
+    # Base domain common to all searches
+    base_domain = [
+        ('parent_state', '=', 'posted'),
+        ('reconciled', '=', False),
+        ('company_id', '=', env.company.id),
+        ('display_type', 'not in', ('line_section', 'line_subsection', 'line_note')),
+        ('statement_line_id', '=', False),
+    ]
+
+    # --- PRIORITY 1: Outstanding payments/receipts on bank journal accounts ---
+    # These are registered payments waiting to be matched to bank lines.
+    # For incoming bank lines → look for outstanding receipts (credit on outstanding account)
+    # For outgoing bank lines → look for outstanding payments (debit on outstanding account)
+    outstanding_acct_ids = env['account.account'].search([
+        ('name', 'ilike', 'outstanding'),
+        ('company_ids', 'in', env.company.id),
+    ]).ids
+    outstanding_amls = AML
+    if outstanding_acct_ids:
+        os_domain = base_domain + [('account_id', 'in', outstanding_acct_ids)]
+        if is_incoming:
+            os_domain.append(('amount_residual', '>', 0))  # Debit residual on outstanding receipts
+        else:
+            os_domain.append(('amount_residual', '<', 0))  # Credit residual on outstanding payments
+        if partner:
+            outstanding_amls = AML.search(os_domain + [('partner_id', '=', partner.id)], limit=30)
+            if not outstanding_amls:
+                outstanding_amls = AML.search(os_domain, limit=30)
+        else:
+            outstanding_amls = AML.search(os_domain, limit=30)
+
+    # --- PRIORITY 2: Open invoices/bills (receivable/payable accounts) ---
+    inv_domain = list(base_domain)
+    if is_incoming:
+        inv_domain.append(('account_id.account_type', '=', 'asset_receivable'))
+        inv_domain.append(('amount_residual', '>', 0))
+    else:
+        inv_domain.append(('account_id.account_type', '=', 'liability_payable'))
+        inv_domain.append(('amount_residual', '<', 0))
+    inv_domain.append(('date', '>=', str(line_date - td(days=90))))
+    inv_domain.append(('date', '<=', str(line_date + td(days=30))))
+
+    invoice_amls = AML
+    if partner:
+        invoice_amls = AML.search(inv_domain + [('partner_id', '=', partner.id)], limit=30)
+        if not invoice_amls:
+            invoice_amls = AML.search(inv_domain, limit=30)
+    else:
+        invoice_amls = AML.search(inv_domain, limit=30)
+
+    # Merge: outstanding payments first (priority), then invoices/bills
+    combined = outstanding_amls | invoice_amls
+
+    # Score and format candidates
+    outstanding_ids = set(outstanding_amls.ids) if outstanding_amls else set()
+    candidates = []
+    seen_ids = set()
+    for aml in combined:
+        if aml.id in seen_ids:
+            continue
+        seen_ids.add(aml.id)
+        residual = abs(aml.amount_residual)
+        score = 0
+        reasons = []
+        is_payment = aml.id in outstanding_ids
+
+        # Source type: payment entries get a boost (preferred match)
+        if is_payment:
+            score += 15
+            reasons.append('payment')
+
+        # Amount scoring
+        if abs(residual - bank_amount) < 0.01:
+            score += 40
+            reasons.append('exact amount')
+        elif residual <= bank_amount * 1.05:
+            score += 20
+            reasons.append('close amount')
+
+        # Partner scoring
+        if partner and aml.partner_id.id == partner.id:
+            score += 25
+            reasons.append('partner')
+        elif partner and aml.partner_id and partner.name and aml.partner_id.name:
+            p1_words = set(partner.name.upper().split())
+            p2_words = set(aml.partner_id.name.upper().split())
+            if p1_words & p2_words:
+                score += 10
+                reasons.append('partial partner')
+
+        # Date proximity scoring
+        days_apart = abs((aml.date - line_date).days)
+        if days_apart <= 3:
+            score += 15
+            reasons.append(f'{days_apart}d')
+        elif days_apart <= 7:
+            score += 10
+        elif days_apart <= 14:
+            score += 5
+
+        # Reference matching
+        if line.payment_ref and aml.move_id.ref:
+            if any(w.upper() in (aml.move_id.ref or '').upper()
+                   for w in line.payment_ref.split() if len(w) > 3):
+                score += 10
+                reasons.append('ref match')
+
+        # Determine entry type label
+        entry_type = 'payment' if is_payment else 'invoice'
+        if aml.move_id.move_type == 'in_invoice':
+            entry_type = 'bill'
+        elif aml.move_id.move_type == 'out_invoice':
+            entry_type = 'invoice'
+        elif aml.move_id.move_type in ('in_refund', 'out_refund'):
+            entry_type = 'credit note'
+        elif aml.payment_id:
+            entry_type = 'payment'
+
+        candidates.append({
+            'aml_id': aml.id,
+            'move_id': aml.move_id.id,
+            'name': aml.move_id.name or '',
+            'ref': aml.move_id.ref or '',
+            'partner': aml.partner_id.name if aml.partner_id else '',
+            'partner_id': aml.partner_id.id if aml.partner_id else None,
+            'date': str(aml.date),
+            'amount_total': abs(aml.balance),
+            'amount_residual': residual,
+            'account': aml.account_id.code if hasattr(aml.account_id, 'code') else '',
+            'type': entry_type,
+            'score': score,
+            'reasons': ', '.join(reasons) if reasons else '',
+        })
+
+    # Sort by score descending
+    candidates.sort(key=lambda c: -c['score'])
+
+    # Find best subset-sum combination
+    best_combo_ids, best_combo_total = _find_best_subset(candidates, bank_amount)
+
+    # Mark which candidates are in the best combination
+    for c in candidates:
+        c['in_best_combo'] = c['aml_id'] in best_combo_ids
+
+    return {
+        'bank_line': {
+            'id': line.id,
+            'date': str(line_date),
+            'ref': line.payment_ref or '',
+            'amount': line.amount,
+            'abs_amount': bank_amount,
+            'journal': line.journal_id.name,
+            'partner': partner.name if partner else '',
+            'partner_id': partner.id if partner else None,
+            'direction': 'incoming' if is_incoming else 'outgoing',
+        },
+        'candidates': candidates[:20],
+        'best_combination': best_combo_ids,
+        'best_combination_total': best_combo_total,
+        'is_exact_match': abs(best_combo_total - bank_amount) < 0.01,
+        'count': len(candidates),
+    }
+
+
+def search_matching_entries(env, params):
+    """Search open journal items by query (invoice/bill number, amount, or partner name).
+    Used by the reconciliation table search bar via direct RPC."""
+    query = (params.get('query') or '').strip()
+    line_id = params.get('statement_line_id')
+    if not query:
+        return {'candidates': []}
+
+    AML = env['account.move.line'].sudo()
+
+    # Search across receivable, payable, AND outstanding accounts
+    outstanding_acct_ids = env['account.account'].search([
+        ('name', 'ilike', 'outstanding'),
+        ('company_ids', 'in', env.company.id),
+    ]).ids
+    domain = [
+        ('parent_state', '=', 'posted'),
+        ('reconciled', '=', False),
+        ('company_id', '=', env.company.id),
+        ('display_type', 'not in', ('line_section', 'line_subsection', 'line_note')),
+        '|',
+        ('account_id.account_type', 'in', ('asset_receivable', 'liability_payable')),
+        ('account_id', 'in', outstanding_acct_ids),
+    ]
+
+    # Try as amount first
+    try:
+        amount = float(query.replace('$', '').replace(',', ''))
+        amount_domain = domain + [
+            '|',
+            '&', ('amount_residual', '>=', amount - 0.50), ('amount_residual', '<=', amount + 0.50),
+            '&', ('amount_residual', '>=', -amount - 0.50), ('amount_residual', '<=', -amount + 0.50),
+        ]
+        amls = AML.search(amount_domain, limit=15)
+        if amls:
+            return {'candidates': _format_aml_candidates(amls)}
+    except ValueError:
+        pass
+
+    # Search by move name (invoice/bill number)
+    name_amls = AML.search(domain + [('move_id.name', 'ilike', query)], limit=15)
+    if name_amls:
+        return {'candidates': _format_aml_candidates(name_amls)}
+
+    # Search by move ref
+    ref_amls = AML.search(domain + [('move_id.ref', 'ilike', query)], limit=15)
+    if ref_amls:
+        return {'candidates': _format_aml_candidates(ref_amls)}
+
+    # Search by partner name
+    partner_amls = AML.search(domain + [('partner_id.name', 'ilike', query)], limit=15)
+    return {'candidates': _format_aml_candidates(partner_amls)}
+
+
+def _format_aml_candidates(amls):
+    """Format AMLs as candidate dicts for the reconciliation table."""
+    return [{
+        'aml_id': aml.id,
+        'move_id': aml.move_id.id,
+        'name': aml.move_id.name or '',
+        'ref': aml.move_id.ref or '',
+        'partner': aml.partner_id.name if aml.partner_id else '',
+        'partner_id': aml.partner_id.id if aml.partner_id else None,
+        'date': str(aml.date),
+        'amount_total': abs(aml.balance),
+        'amount_residual': abs(aml.amount_residual),
+        'account': aml.account_id.code if hasattr(aml.account_id, 'code') else '',
+        'score': 0,
+        'reasons': 'manual search',
+        'in_best_combo': False,
+    } for aml in amls]
+
+
+TOOLS = {
+    'get_unreconciled_bank_lines': get_unreconciled_bank_lines,
+    'get_unreconciled_receipts': get_unreconciled_receipts,
+    'match_bank_line_to_payments': match_bank_line_to_payments,
+    'auto_reconcile_bank_lines': auto_reconcile_bank_lines,
+    'apply_reconcile_model': apply_reconcile_model,
+    'unmatch_bank_line': unmatch_bank_line,
+    'get_reconcile_suggestions': get_reconcile_suggestions,
+    'sum_payments_by_date': sum_payments_by_date,
+    'get_bank_line_details': get_bank_line_details,
+    'check_recurring_pattern': check_recurring_pattern,
+    'match_internal_transfers': match_internal_transfers,
+    'find_unreconciled_cheques': find_unreconciled_cheques,
+    'reconcile_payroll_cheques': reconcile_payroll_cheques,
+    'suggest_bank_line_matches': suggest_bank_line_matches,
+    'search_matching_entries': search_matching_entries,
+}

fusion_accounting_ai/services/tools/hst_management.py

@@ -0,0 +1,299 @@
+import logging
+
+_logger = logging.getLogger(__name__)
+
+
+def calculate_hst_balance(env, params):
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+    base_domain = [
+        ('parent_state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ]
+    if date_from:
+        base_domain.append(('date', '>=', date_from))
+    if date_to:
+        base_domain.append(('date', '<=', date_to))
+
+    # Odoo 19 Enterprise: account.account may not have company_id field
+    # (shared chart of accounts). Use try/except to handle both cases.
+    try:
+        collected_accounts = env['account.account'].search([
+            ('code', '=like', '2005%'), ('company_ids', 'in', env.company.id),
+        ])
+        itc_accounts = env['account.account'].search([
+            ('code', '=like', '2006%'), ('company_ids', 'in', env.company.id),
+        ])
+    except Exception:
+        collected_accounts = env['account.account'].search([
+            ('code', '=like', '2005%'),
+        ])
+        itc_accounts = env['account.account'].search([
+            ('code', '=like', '2006%'),
+        ])
+
+    collected_lines = env['account.move.line'].search(
+        base_domain + [('account_id', 'in', collected_accounts.ids)]
+    )
+    itc_lines = env['account.move.line'].search(
+        base_domain + [('account_id', 'in', itc_accounts.ids)]
+    )
+
+    hst_collected = abs(sum(l.balance for l in collected_lines))
+    itcs = abs(sum(l.balance for l in itc_lines))
+
+    return {
+        'hst_collected': hst_collected,
+        'input_tax_credits': itcs,
+        'net_hst': hst_collected - itcs,
+        'status': 'owing' if (hst_collected - itcs) > 0 else 'refund',
+        'period': f'{date_from or "all"} to {date_to or "now"}',
+    }
+
+
+def get_tax_report(env, params):
+    report_ref = params.get('report_ref', 'account.generic_tax_report')
+    try:
+        report = env.ref(report_ref)
+    except Exception:
+        return {'error': f'Report not found: {report_ref}'}
+    options = report.get_options({
+        'date': {
+            'date_from': params.get('date_from', ''),
+            'date_to': params.get('date_to', ''),
+        }
+    })
+    lines = report._get_lines(options)
+    return {
+        'report_name': report.name,
+        'lines': [{
+            'name': l.get('name', ''),
+            'columns': [c.get('no_format', c.get('name', '')) for c in l.get('columns', [])],
+        } for l in lines[:50]],
+    }
+
+
+def find_missing_tax_invoices(env, params):
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+    domain = [
+        ('move_type', 'in', ('out_invoice', 'out_refund')),
+        ('state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ]
+    if date_from:
+        domain.append(('date', '>=', date_from))
+    if date_to:
+        domain.append(('date', '<=', date_to))
+
+    invoices = env['account.move'].search(domain)
+    missing = invoices.filtered(
+        lambda inv: not any(line.tax_ids for line in inv.invoice_line_ids)
+    )
+    return {
+        'total_invoices': len(invoices),
+        'missing_tax_count': len(missing),
+        'invoices': [{
+            'id': inv.id,
+            'name': inv.name,
+            'partner': inv.partner_id.name if inv.partner_id else '',
+            'amount_total': inv.amount_total,
+            'date': str(inv.date),
+        } for inv in missing[:30]],
+    }
+
+
+def find_missing_itc_bills(env, params):
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+    domain = [
+        ('move_type', 'in', ('in_invoice', 'in_refund')),
+        ('state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ]
+    if date_from:
+        domain.append(('date', '>=', date_from))
+    if date_to:
+        domain.append(('date', '<=', date_to))
+
+    bills = env['account.move'].search(domain)
+    missing = bills.filtered(
+        lambda b: not any(line.tax_ids for line in b.invoice_line_ids)
+    )
+    return {
+        'total_bills': len(bills),
+        'missing_itc_count': len(missing),
+        'bills': [{
+            'id': b.id,
+            'name': b.name,
+            'partner': b.partner_id.name if b.partner_id else '',
+            'amount_total': b.amount_total,
+            'date': str(b.date),
+        } for b in missing[:30]],
+    }
+
+
+def get_tax_return_status(env, params):
+    try:
+        AccountReturn = env['account.return']
+    except KeyError:
+        return {'error': 'Tax return model (account.return) is not available. The account_tax_report or related Enterprise module may not be installed.'}
+    returns = AccountReturn.search([
+        ('company_id', '=', env.company.id),
+    ], order='date_start desc', limit=10)
+    return {
+        'returns': [{
+            'id': r.id,
+            'name': r.display_name,
+            'date_start': str(r.date_start) if hasattr(r, 'date_start') else '',
+            'date_end': str(r.date_end) if hasattr(r, 'date_end') else '',
+            'state': r.state if hasattr(r, 'state') else '',
+        } for r in returns],
+    }
+
+
+def generate_tax_return(env, params):
+    try:
+        AccountReturn = env['account.return']
+    except KeyError:
+        return {'error': 'Tax return model (account.return) is not available.'}
+    try:
+        AccountReturn._generate_or_refresh_all_returns(
+            company=env.company
+        )
+        return {'status': 'generated', 'message': 'Tax returns refreshed successfully.'}
+    except Exception as e:
+        return {'error': str(e)}
+
+
+def validate_tax_return(env, params):
+    try:
+        AccountReturn = env['account.return']
+    except KeyError:
+        return {'error': 'Tax return model (account.return) is not available.'}
+    return_id = int(params['return_id'])
+    tax_return = AccountReturn.browse(return_id)
+    if not tax_return.exists():
+        return {'error': 'Tax return not found'}
+    try:
+        tax_return.action_validate()
+        return {'status': 'validated', 'return_id': return_id}
+    except Exception as e:
+        return {'error': str(e)}
+
+
+def create_expense_entry(env, params):
+    """[Tier 3] Create a direct GL expense entry in the Misc journal with optional HST split.
+    This is the 'old school' way of recording expenses without a formal vendor bill.
+    Requires user approval before execution."""
+    date = params.get('date', str(env['account.move']._fields['date'].default(env['account.move'])))
+    description = params.get('description', 'Expense')
+    expense_account_id = int(params['expense_account_id'])
+    amount = abs(float(params['amount']))
+    has_hst = params.get('has_hst', False)
+    bank_journal_id = int(params.get('bank_journal_id', 0))
+
+    # Find the MISC journal
+    misc_journal = env['account.journal'].search([
+        ('code', '=', 'MISC'), ('company_id', '=', env.company.id),
+    ], limit=1)
+    if not misc_journal:
+        return {'error': 'Miscellaneous Operations journal (MISC) not found'}
+
+    expense_account = env['account.account'].browse(expense_account_id)
+    if not expense_account.exists():
+        return {'error': f'Expense account not found: {expense_account_id}'}
+
+    # Determine credit account (bank outstanding or AP)
+    credit_account = None
+    if bank_journal_id:
+        bank_journal = env['account.journal'].browse(bank_journal_id)
+        if bank_journal.exists():
+            # Use the bank journal's default debit/credit account
+            credit_account = (bank_journal.default_account_id
+                              or bank_journal.company_id.account_journal_payment_credit_account_id)
+    if not credit_account:
+        # Fallback to AP account
+        credit_account = env['account.account'].search([
+            ('account_type', '=', 'liability_payable'),
+            ('company_ids', 'in', env.company.id),
+        ], limit=1)
+
+    if not credit_account:
+        return {'error': 'Could not determine credit account for the expense entry'}
+
+    line_ids = []
+    if has_hst:
+        # Split: net expense + 13% HST ITC
+        hst_rate = 0.13
+        net_amount = round(amount / (1 + hst_rate), 2)
+        hst_amount = round(amount - net_amount, 2)
+
+        # Find HST ITC account (2006%)
+        itc_account = env['account.account'].search([
+            ('code', '=like', '2006%'),
+        ], limit=1)
+        if not itc_account:
+            # Fallback: use the HST purchase tax account
+            hst_tax = env['account.tax'].search([
+                ('type_tax_use', '=', 'purchase'), ('amount', '=', 13.0),
+                ('company_id', '=', env.company.id),
+            ], limit=1)
+            if hst_tax and hst_tax.invoice_repartition_line_ids:
+                for rep in hst_tax.invoice_repartition_line_ids:
+                    if rep.repartition_type == 'tax' and rep.account_id:
+                        itc_account = rep.account_id
+                        break
+        if not itc_account:
+            return {'error': 'HST ITC account (2006) not found'}
+
+        line_ids = [
+            (0, 0, {'name': description, 'account_id': expense_account_id,
+                     'debit': net_amount, 'credit': 0.0}),
+            (0, 0, {'name': f'HST ITC - {description}', 'account_id': itc_account.id,
+                     'debit': hst_amount, 'credit': 0.0}),
+            (0, 0, {'name': description, 'account_id': credit_account.id,
+                     'debit': 0.0, 'credit': amount}),
+        ]
+    else:
+        # Simple: debit expense / credit bank
+        line_ids = [
+            (0, 0, {'name': description, 'account_id': expense_account_id,
+                     'debit': amount, 'credit': 0.0}),
+            (0, 0, {'name': description, 'account_id': credit_account.id,
+                     'debit': 0.0, 'credit': amount}),
+        ]
+
+    try:
+        move = env['account.move'].create({
+            'move_type': 'entry',
+            'journal_id': misc_journal.id,
+            'date': date,
+            'ref': description,
+            'line_ids': line_ids,
+            'company_id': env.company.id,
+        })
+        move.action_post()
+        return {
+            'status': 'posted',
+            'move_id': move.id,
+            'move_name': move.name,
+            'amount': amount,
+            'has_hst': has_hst,
+            'hst_amount': round(amount - amount / 1.13, 2) if has_hst else 0.0,
+        }
+    except Exception as e:
+        _logger.error("Failed to create expense entry: %s", e)
+        return {'error': str(e)}
+
+
+TOOLS = {
+    'calculate_hst_balance': calculate_hst_balance,
+    'get_tax_report': get_tax_report,
+    'find_missing_tax_invoices': find_missing_tax_invoices,
+    'find_missing_itc_bills': find_missing_itc_bills,
+    'get_tax_return_status': get_tax_return_status,
+    'generate_tax_return': generate_tax_return,
+    'validate_tax_return': validate_tax_return,
+    'create_expense_entry': create_expense_entry,
+}

fusion_accounting_ai/services/tools/inventory.py

@@ -0,0 +1,113 @@
+import logging
+from odoo import fields
+
+_logger = logging.getLogger(__name__)
+
+
+def get_stock_valuation(env, params):
+    accounts = env['account.account'].search([
+        ('code', '=like', '1069%'),
+        ('company_ids', 'in', env.company.id),
+    ])
+    result = []
+    for acct in accounts:
+        balance = sum(env['account.move.line'].search([
+            ('account_id', '=', acct.id),
+            ('parent_state', '=', 'posted'),
+        ]).mapped('balance'))
+        result.append({'code': acct.code, 'name': acct.name, 'balance': balance})
+    return {'accounts': result, 'total': sum(r['balance'] for r in result)}
+
+
+def get_price_differences(env, params):
+    accounts = env['account.account'].search([
+        ('code', '=like', '5010%'),
+        ('company_ids', 'in', env.company.id),
+    ])
+    domain = [
+        ('account_id', 'in', accounts.ids),
+        ('parent_state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ]
+    if params.get('date_from'):
+        domain.append(('date', '>=', params['date_from']))
+    if params.get('date_to'):
+        domain.append(('date', '<=', params['date_to']))
+    lines = env['account.move.line'].search(domain, order='date desc', limit=50)
+    return {
+        'total': sum(l.balance for l in lines),
+        'entries': [{
+            'id': l.id, 'date': str(l.date),
+            'move': l.move_id.name, 'amount': l.balance,
+            'partner': l.partner_id.name if l.partner_id else '',
+        } for l in lines],
+    }
+
+
+def get_cogs_ratio_by_category(env, params):
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+    base_domain = [
+        ('parent_state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ]
+    if date_from:
+        base_domain.append(('date', '>=', date_from))
+    if date_to:
+        base_domain.append(('date', '<=', date_to))
+
+    revenue_lines = env['account.move.line'].search(
+        base_domain + [('account_id.account_type', '=', 'income')]
+    )
+    cogs_lines = env['account.move.line'].search(
+        base_domain + [('account_id.account_type', '=', 'expense_direct_cost')]
+    )
+    revenue = abs(sum(l.balance for l in revenue_lines))
+    cogs = abs(sum(l.balance for l in cogs_lines))
+    ratio = (cogs / revenue * 100) if revenue else 0
+    return {'revenue': revenue, 'cogs': cogs, 'ratio_pct': round(ratio, 2)}
+
+
+def find_unusual_adjustments(env, params):
+    threshold = float(params.get('threshold', 1000))
+    domain = [
+        ('parent_state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+        ('account_id.account_type', '=', 'expense_direct_cost'),
+    ]
+    lines = env['account.move.line'].search(domain)
+    unusual = lines.filtered(lambda l: abs(l.balance) > threshold)
+    return {
+        'count': len(unusual),
+        'adjustments': [{
+            'id': l.id, 'date': str(l.date), 'move': l.move_id.name,
+            'amount': l.balance, 'name': l.name or '',
+        } for l in unusual[:20]],
+    }
+
+
+def get_inventory_turnover(env, params):
+    from .reporting import get_profit_loss
+    pl = get_profit_loss(env, params)
+    stock = get_stock_valuation(env, params)
+    avg_inventory = stock.get('total', 0)
+    cogs = 0
+    for line in pl.get('lines', []):
+        if 'cost' in line.get('name', '').lower():
+            cols = line.get('columns', [])
+            if cols:
+                try:
+                    cogs = float(cols[0])
+                except (ValueError, TypeError):
+                    pass
+    turnover = (cogs / avg_inventory) if avg_inventory else 0
+    return {'cogs': cogs, 'avg_inventory': avg_inventory, 'turnover': round(turnover, 2)}
+
+
+TOOLS = {
+    'get_stock_valuation': get_stock_valuation,
+    'get_price_differences': get_price_differences,
+    'get_cogs_ratio_by_category': get_cogs_ratio_by_category,
+    'find_unusual_adjustments': find_unusual_adjustments,
+    'get_inventory_turnover': get_inventory_turnover,
+}

fusion_accounting_ai/services/tools/journal_review.py

@@ -0,0 +1,220 @@
+import logging
+from odoo import fields
+
+_logger = logging.getLogger(__name__)
+
+ACCOUNT_TYPE_EXPECTED_DIRECTION = {
+    'asset_receivable': 'debit',
+    'asset_cash': 'debit',
+    'asset_current': 'debit',
+    'asset_non_current': 'debit',
+    'asset_prepayments': 'debit',
+    'asset_fixed': 'debit',
+    'liability_payable': 'credit',
+    'liability_credit_card': 'credit',
+    'liability_current': 'credit',
+    'liability_non_current': 'credit',
+    'equity': 'credit',
+    'equity_unaffected': 'credit',
+    'income': 'credit',
+    'income_other': 'credit',
+    'expense': 'debit',
+    'expense_depreciation': 'debit',
+    'expense_direct_cost': 'debit',
+    'off_balance': None,
+}
+
+
+def find_wrong_direction_balances(env, params):
+    balance_data = env['account.move.line'].read_group(
+        [('parent_state', '=', 'posted'), ('company_id', '=', env.company.id)],
+        ['balance:sum'], ['account_id'],
+    )
+    acct_ids = [row['account_id'][0] for row in balance_data if row.get('account_id')]
+    acct_map = {}
+    if acct_ids:
+        for acct in env['account.account'].browse(acct_ids):
+            acct_map[acct.id] = acct
+
+    issues = []
+    for row in balance_data:
+        if not row.get('account_id'):
+            continue
+        acct = acct_map.get(row['account_id'][0])
+        if not acct:
+            continue
+        expected = ACCOUNT_TYPE_EXPECTED_DIRECTION.get(acct.account_type)
+        if not expected:
+            continue
+        balance = row.get('balance', 0) or 0
+        if (expected == 'debit' and balance < -0.01) or (expected == 'credit' and balance > 0.01):
+            issues.append({
+                'account_id': acct.id,
+                'code': acct.code,
+                'name': acct.name,
+                'balance': balance,
+                'expected': expected,
+                'actual': 'credit' if balance < 0 else 'debit',
+            })
+    return {'count': len(issues), 'issues': issues}
+
+
+def find_duplicate_entries(env, params):
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+    domain = [
+        ('state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ]
+    if date_from:
+        domain.append(('date', '>=', date_from))
+    if date_to:
+        domain.append(('date', '<=', date_to))
+    moves = env['account.move'].search(domain, order='partner_id, amount_total, date')
+
+    duplicates = []
+    prev = None
+    for move in moves:
+        if prev and (
+            prev.partner_id == move.partner_id and prev.partner_id
+            and abs(prev.amount_total - move.amount_total) < 0.01
+            and prev.date == move.date
+            and prev.journal_id == move.journal_id
+        ):
+            duplicates.append({
+                'entry_1': {'id': prev.id, 'name': prev.name},
+                'entry_2': {'id': move.id, 'name': move.name},
+                'partner': move.partner_id.name,
+                'amount': move.amount_total,
+                'date': str(move.date),
+            })
+        prev = move
+    return {'count': len(duplicates), 'duplicates': duplicates[:20]}
+
+
+def find_wrong_account_entries(env, params):
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+    domain = [
+        ('parent_state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ]
+    if date_from:
+        domain.append(('date', '>=', date_from))
+    if date_to:
+        domain.append(('date', '<=', date_to))
+
+    issues = []
+    tax_accounts = env['account.account'].search([
+        ('account_type', 'in', ('liability_current', 'asset_current')),
+        ('code', '=like', '2005%'),
+        ('company_ids', 'in', env.company.id),
+    ])
+    if tax_accounts:
+        revenue_on_tax = env['account.move.line'].search(
+            domain + [
+                ('account_id', 'in', tax_accounts.ids),
+                ('product_id', '!=', False),
+            ]
+        )
+        for line in revenue_on_tax[:20]:
+            issues.append({
+                'id': line.id,
+                'move': line.move_id.name,
+                'account': f'{line.account_id.code} {line.account_id.name}',
+                'product': line.product_id.name,
+                'amount': line.balance,
+                'issue': 'Product line on tax account',
+            })
+    return {'count': len(issues), 'issues': issues}
+
+
+def find_sequence_gaps(env, params):
+    moves = env['account.move'].search([
+        ('state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+        ('made_sequence_gap', '=', True),
+    ], order='date desc', limit=50)
+    return {
+        'count': len(moves),
+        'gaps': [{
+            'id': m.id,
+            'name': m.name,
+            'date': str(m.date),
+            'journal': m.journal_id.name,
+        } for m in moves],
+    }
+
+
+def find_draft_entries(env, params):
+    min_age_days = int(params.get('min_age_days', 30))
+    from datetime import timedelta
+    cutoff = fields.Date.today() - timedelta(days=min_age_days)
+    drafts = env['account.move'].search([
+        ('state', '=', 'draft'),
+        ('date', '<=', cutoff),
+        ('company_id', '=', env.company.id),
+    ], order='date asc', limit=50)
+    return {
+        'count': len(drafts),
+        'entries': [{
+            'id': d.id,
+            'name': d.name or 'Draft',
+            'date': str(d.date),
+            'journal': d.journal_id.name,
+            'amount': d.amount_total,
+            'partner': d.partner_id.name if d.partner_id else '',
+        } for d in drafts],
+    }
+
+
+def find_unreconciled_suspense(env, params):
+    suspense_accounts = env['account.account'].search([
+        ('code', '=like', '999%'),
+        ('company_ids', 'in', env.company.id),
+    ])
+    issues = []
+    for acct in suspense_accounts:
+        balance = sum(env['account.move.line'].search([
+            ('account_id', '=', acct.id),
+            ('parent_state', '=', 'posted'),
+        ]).mapped('balance'))
+        if abs(balance) > 0.01:
+            issues.append({
+                'account_id': acct.id,
+                'code': acct.code,
+                'name': acct.name,
+                'balance': balance,
+            })
+    return {'count': len(issues), 'accounts': issues}
+
+
+def verify_reconciliation_integrity(env, params):
+    partials = env['account.partial.reconcile'].search([
+        ('company_id', '=', env.company.id),
+    ], limit=500)
+    issues = []
+    for p in partials:
+        debit_ok = p.debit_move_id.reconciled or abs(p.debit_move_id.amount_residual) < 0.01
+        credit_ok = p.credit_move_id.reconciled or abs(p.credit_move_id.amount_residual) < 0.01
+        if not debit_ok and not credit_ok:
+            issues.append({
+                'id': p.id,
+                'debit_move': p.debit_move_id.move_id.name,
+                'credit_move': p.credit_move_id.move_id.name,
+                'amount': p.amount,
+                'debit_residual': p.debit_move_id.amount_residual,
+                'credit_residual': p.credit_move_id.amount_residual,
+            })
+    return {'count': len(issues), 'issues': issues[:20]}
+
+
+TOOLS = {
+    'find_wrong_direction_balances': find_wrong_direction_balances,
+    'find_duplicate_entries': find_duplicate_entries,
+    'find_wrong_account_entries': find_wrong_account_entries,
+    'find_sequence_gaps': find_sequence_gaps,
+    'find_draft_entries': find_draft_entries,
+    'find_unreconciled_suspense': find_unreconciled_suspense,
+    'verify_reconciliation_integrity': verify_reconciliation_integrity,
+}

fusion_accounting_ai/services/tools/month_end.py

@@ -0,0 +1,130 @@
+import logging
+from odoo import fields
+
+_logger = logging.getLogger(__name__)
+
+
+def get_close_checklist(env, params):
+    from .bank_reconciliation import get_unreconciled_bank_lines
+    from .journal_review import find_draft_entries, find_sequence_gaps
+    from .hst_management import calculate_hst_balance
+
+    period = params.get('period', str(fields.Date.today())[:7])
+    date_from = f'{period}-01'
+    import calendar
+    year, month = int(period[:4]), int(period[5:7])
+    last_day = calendar.monthrange(year, month)[1]
+    date_to = f'{period}-{last_day:02d}'
+
+    p = {'date_from': date_from, 'date_to': date_to}
+
+    bank = get_unreconciled_bank_lines(env, p)
+    drafts = find_draft_entries(env, {'min_age_days': '0'})
+    gaps = find_sequence_gaps(env, p)
+    hst = calculate_hst_balance(env, p)
+
+    checklist = [
+        {'item': 'Bank Reconciliation', 'status': 'ok' if bank['count'] == 0 else 'attention', 'detail': f"{bank['count']} unreconciled lines"},
+        {'item': 'Draft Entries', 'status': 'ok' if drafts['count'] == 0 else 'attention', 'detail': f"{drafts['count']} draft entries"},
+        {'item': 'Sequence Gaps', 'status': 'ok' if gaps['count'] == 0 else 'warning', 'detail': f"{gaps['count']} gaps found"},
+        {'item': 'HST Balance', 'status': 'info', 'detail': f"Net HST: ${hst['net_hst']:.2f}"},
+    ]
+    return {'period': period, 'checklist': checklist}
+
+
+def get_unreconciled_counts(env, params):
+    accounts = env['account.account'].search([
+        ('reconcile', '=', True),
+        ('company_ids', 'in', env.company.id),
+    ])
+    result = []
+    for acct in accounts:
+        count = env['account.move.line'].search_count([
+            ('account_id', '=', acct.id),
+            ('reconciled', '=', False),
+            ('parent_state', '=', 'posted'),
+        ])
+        if count > 0:
+            result.append({
+                'account_id': acct.id,
+                'code': acct.code,
+                'name': acct.name,
+                'unreconciled_count': count,
+            })
+    return {'accounts': sorted(result, key=lambda x: -x['unreconciled_count'])}
+
+
+def find_entries_in_locked_period(env, params):
+    company = env.company
+    lock_date = company.fiscalyear_lock_date
+    if not lock_date:
+        return {'status': 'no_lock_date', 'entries': []}
+    entries = env['account.move'].search([
+        ('date', '<=', lock_date),
+        ('state', '=', 'draft'),
+        ('company_id', '=', company.id),
+    ])
+    return {
+        'lock_date': str(lock_date),
+        'count': len(entries),
+        'entries': [{'id': e.id, 'name': e.name, 'date': str(e.date)} for e in entries[:20]],
+    }
+
+
+def get_accrual_status(env, params):
+    accrual_codes = params.get('account_codes', ['2100', '2110', '2120'])
+    result = []
+    for code in accrual_codes:
+        accounts = env['account.account'].search([
+            ('code', '=like', f'{code}%'),
+            ('company_ids', 'in', env.company.id),
+        ])
+        for acct in accounts:
+            balance = sum(env['account.move.line'].search([
+                ('account_id', '=', acct.id),
+                ('parent_state', '=', 'posted'),
+            ]).mapped('balance'))
+            result.append({'code': acct.code, 'name': acct.name, 'balance': balance})
+    return {'accruals': result}
+
+
+def run_hash_integrity_check(env, params):
+    try:
+        result = env.company._check_hash_integrity()
+        return {
+            'status': 'completed',
+            'results': result.get('results', []),
+            'printing_date': result.get('printing_date', ''),
+        }
+    except Exception as e:
+        return {'error': str(e)}
+
+
+def get_period_summary(env, params):
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+    try:
+        report = env.ref('account_reports.trial_balance_report')
+    except Exception:
+        report = env.ref('account.trial_balance_report', raise_if_not_found=False)
+    if not report:
+        return {'error': 'Trial balance report not found'}
+    options = report.get_options({'date': {'date_from': date_from, 'date_to': date_to}})
+    lines = report._get_lines(options)
+    return {
+        'period': f'{date_from} to {date_to}',
+        'lines': [{
+            'name': l.get('name', ''),
+            'columns': [c.get('no_format', c.get('name', '')) for c in l.get('columns', [])],
+        } for l in lines[:100]],
+    }
+
+
+TOOLS = {
+    'get_close_checklist': get_close_checklist,
+    'get_unreconciled_counts': get_unreconciled_counts,
+    'find_entries_in_locked_period': find_entries_in_locked_period,
+    'get_accrual_status': get_accrual_status,
+    'run_hash_integrity_check': run_hash_integrity_check,
+    'get_period_summary': get_period_summary,
+}

fusion_accounting_ai/services/tools/payroll.py

@@ -0,0 +1,256 @@
+import logging
+from odoo import fields
+
+_logger = logging.getLogger(__name__)
+
+
+def get_payroll_entries(env, params):
+    payroll_journals = env['account.journal'].search([
+        ('name', 'ilike', 'payroll'),
+        ('company_id', '=', env.company.id),
+    ])
+    if not payroll_journals and params.get('journal_id'):
+        payroll_journals = env['account.journal'].browse(int(params['journal_id']))
+    domain = [
+        ('journal_id', 'in', payroll_journals.ids),
+        ('state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ]
+    if params.get('date_from'):
+        domain.append(('date', '>=', params['date_from']))
+    if params.get('date_to'):
+        domain.append(('date', '<=', params['date_to']))
+    entries = env['account.move'].search(domain, order='date desc', limit=50)
+    return {
+        'count': len(entries),
+        'entries': [{
+            'id': e.id, 'name': e.name, 'date': str(e.date),
+            'amount': e.amount_total, 'ref': e.ref or '',
+        } for e in entries],
+    }
+
+
+def compare_payroll_to_bank(env, params):
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+    if not date_from or not date_to:
+        return {'error': 'date_from and date_to are required'}
+    payroll_journals = env['account.journal'].search([
+        ('name', 'ilike', 'payroll'), ('company_id', '=', env.company.id),
+    ])
+    payroll_entries = env['account.move'].search([
+        ('journal_id', 'in', payroll_journals.ids),
+        ('state', '=', 'posted'),
+        ('date', '>=', date_from), ('date', '<=', date_to),
+    ])
+    bank_lines = env['account.bank.statement.line'].search([
+        ('date', '>=', date_from), ('date', '<=', date_to),
+        ('company_id', '=', env.company.id),
+    ])
+    payroll_total = sum(e.amount_total for e in payroll_entries)
+    bank_payroll = sum(abs(l.amount) for l in bank_lines if 'payroll' in (l.payment_ref or '').lower())
+    return {
+        'payroll_journal_total': payroll_total,
+        'bank_payroll_total': bank_payroll,
+        'difference': payroll_total - bank_payroll,
+    }
+
+
+def verify_source_deductions(env, params):
+    return {
+        'status': 'info',
+        'message': 'Source deduction verification requires CRA rate tables. Use fusion_payroll for full verification.',
+    }
+
+
+def get_cra_remittance_status(env, params):
+    cra_accounts = env['account.account'].search([
+        ('name', 'ilike', 'CRA'),
+        ('company_ids', 'in', env.company.id),
+    ])
+    result = []
+    for acct in cra_accounts:
+        balance = sum(env['account.move.line'].search([
+            ('account_id', '=', acct.id),
+            ('parent_state', '=', 'posted'),
+        ]).mapped('balance'))
+        result.append({'code': acct.code, 'name': acct.name, 'balance': balance})
+    return {'accounts': result}
+
+
+def find_unmatched_payroll_cheques(env, params):
+    bank_lines = env['account.bank.statement.line'].search([
+        ('is_reconciled', '=', False),
+        ('company_id', '=', env.company.id),
+        ('payment_ref', 'ilike', 'cheque'),
+    ])
+    return {
+        'count': len(bank_lines),
+        'cheques': [{
+            'id': l.id, 'date': str(l.date),
+            'ref': l.payment_ref, 'amount': l.amount,
+        } for l in bank_lines[:30]],
+    }
+
+
+def parse_payroll_summary(env, params):
+    import re
+    raw_data = params.get('data', '')
+    if not raw_data:
+        return {'error': 'No payroll data provided'}
+
+    lines = raw_data.strip().split('\n')
+    entries = []
+    totals = {'gross': 0, 'cpp': 0, 'ei': 0, 'tax': 0, 'net': 0}
+
+    for line in lines:
+        amounts = re.findall(r'\$?([\d,]+\.?\d*)', line)
+        if len(amounts) >= 2:
+            name_part = re.sub(r'\$?[\d,]+\.?\d*', '', line).strip(' \t,|-')
+            parsed_amounts = [float(a.replace(',', '')) for a in amounts]
+            entry = {'name': name_part or 'Employee', 'amounts': parsed_amounts}
+            if len(parsed_amounts) >= 5:
+                entry.update({
+                    'gross': parsed_amounts[0],
+                    'cpp': parsed_amounts[1],
+                    'ei': parsed_amounts[2],
+                    'tax': parsed_amounts[3],
+                    'net': parsed_amounts[4] if len(parsed_amounts) > 4 else parsed_amounts[0] - sum(parsed_amounts[1:4]),
+                })
+                for k in ('gross', 'cpp', 'ei', 'tax', 'net'):
+                    totals[k] += entry.get(k, 0)
+            entries.append(entry)
+
+    return {
+        'status': 'parsed',
+        'employee_count': len(entries),
+        'entries': entries,
+        'totals': totals,
+        'raw_lines': len(lines),
+    }
+
+
+def _resolve_account_id(env, val):
+    """Resolve an account code or ID to a valid account ID.
+    Accepts: integer ID, string ID, or account code string like '2201'."""
+    if not val:
+        return False
+    val_str = str(val).strip()
+    # Try as a direct ID first
+    try:
+        acct = env['account.account'].browse(int(val_str))
+        if acct.exists():
+            return acct.id
+    except (ValueError, TypeError):
+        pass
+    # Try as an account code
+    acct = env['account.account'].search([
+        ('code', '=', val_str),
+        ('company_ids', 'in', env.company.id),
+    ], limit=1)
+    if acct:
+        return acct.id
+    return False
+
+
+def create_payroll_journal_entry(env, params):
+    journal_id = int(params['journal_id'])
+    date = params['date']
+    ref = params.get('ref', 'Payroll Entry')
+    lines_data = params['lines']
+
+    # Duplicate check: same journal + date + ref + similar amount
+    total_debit = sum(float(l.get('debit', 0)) for l in lines_data)
+    existing = env['account.move'].search([
+        ('journal_id', '=', journal_id),
+        ('date', '=', date),
+        ('ref', 'ilike', ref[:30]),
+        ('state', 'in', ('draft', 'posted')),
+    ], limit=1)
+    if existing:
+        return {
+            'status': 'duplicate',
+            'error': f'Entry already exists: {existing.name} (ref: {existing.ref}) on {existing.date} '
+                     f'for ${existing.amount_total:,.2f}. Skipping to avoid duplicate.',
+            'existing_move_id': existing.id,
+            'existing_name': existing.name,
+        }
+
+    # Resolve account codes to IDs
+    resolved_lines = []
+    for line in lines_data:
+        account_id = _resolve_account_id(env, line['account_id'])
+        if not account_id:
+            return {'error': f"Account not found: {line['account_id']}. "
+                             f"Provide a valid account code (e.g. '2201') or database ID."}
+        resolved_lines.append((0, 0, {
+            'account_id': account_id,
+            'name': line.get('name', 'Payroll'),
+            'debit': float(line.get('debit', 0)),
+            'credit': float(line.get('credit', 0)),
+            'partner_id': int(line['partner_id']) if line.get('partner_id') else False,
+        }))
+
+    move_vals = {
+        'journal_id': journal_id,
+        'date': date,
+        'ref': ref,
+        'line_ids': resolved_lines,
+    }
+    move = env['account.move'].create(move_vals)
+    return {'status': 'created', 'move_id': move.id, 'name': move.name}
+
+
+def get_payroll_schedule(env, params):
+    return {'status': 'info', 'message': 'Payroll schedule available via fusion_payroll module.'}
+
+
+def match_payroll_cheques(env, params):
+    st_line_id = int(params['statement_line_id'])
+    move_line_ids = [int(x) for x in params['move_line_ids']]
+    st_line = env['account.bank.statement.line'].browse(st_line_id)
+    st_line.set_line_bank_statement_line(move_line_ids)
+    return {'status': 'matched', 'statement_line_id': st_line_id}
+
+
+def verify_payroll_deductions(env, params):
+    return verify_source_deductions(env, params)
+
+
+def get_cra_remittance_due(env, params):
+    return get_cra_remittance_status(env, params)
+
+
+def prepare_cra_payment(env, params):
+    return create_payroll_journal_entry(env, params)
+
+
+def generate_t4(env, params):
+    return {'status': 'info', 'message': 'T4 generation available via fusion_payroll module.'}
+
+
+def generate_roe(env, params):
+    return {'status': 'info', 'message': 'ROE generation available via fusion_payroll module.'}
+
+
+def get_payroll_cost_report(env, params):
+    return get_payroll_entries(env, params)
+
+
+TOOLS = {
+    'get_payroll_entries': get_payroll_entries,
+    'compare_payroll_to_bank': compare_payroll_to_bank,
+    'verify_source_deductions': verify_source_deductions,
+    'get_cra_remittance_status': get_cra_remittance_status,
+    'find_unmatched_payroll_cheques': find_unmatched_payroll_cheques,
+    'parse_payroll_summary': parse_payroll_summary,
+    'create_payroll_journal_entry': create_payroll_journal_entry,
+    'get_payroll_schedule': get_payroll_schedule,
+    'match_payroll_cheques': match_payroll_cheques,
+    'verify_payroll_deductions': verify_payroll_deductions,
+    'get_cra_remittance_due': get_cra_remittance_due,
+    'prepare_cra_payment': prepare_cra_payment,
+    'generate_t4': generate_t4,
+    'generate_roe': generate_roe,
+    'get_payroll_cost_report': get_payroll_cost_report,
+}

fusion_accounting_ai/services/tools/reporting.py

@@ -0,0 +1,285 @@
+import logging
+import base64
+
+_logger = logging.getLogger(__name__)
+
+
+def _get_report(env, ref_id):
+    try:
+        return env.ref(ref_id)
+    except Exception:
+        return None
+
+
+def _run_report(env, report_ref, params):
+    report = _get_report(env, report_ref)
+    if not report:
+        return {'error': f'Report {report_ref} not found'}
+    date_opts = {}
+    if params.get('date_from'):
+        date_opts['date_from'] = params['date_from']
+    if params.get('date_to'):
+        date_opts['date_to'] = params['date_to']
+    options = report.get_options({'date': date_opts} if date_opts else {})
+    lines = report._get_lines(options)
+    return {
+        'report_name': report.name,
+        'lines': [{
+            'name': l.get('name', ''),
+            'level': l.get('level', 0),
+            'columns': [c.get('no_format', c.get('name', '')) for c in l.get('columns', [])],
+        } for l in lines[:100]],
+    }
+
+
+def get_profit_loss(env, params):
+    return _run_report(env, 'account_reports.profit_and_loss', params)
+
+
+def get_balance_sheet(env, params):
+    return _run_report(env, 'account_reports.balance_sheet', params)
+
+
+def get_trial_balance(env, params):
+    return _run_report(env, 'account_reports.trial_balance_report', params)
+
+
+def get_cash_flow(env, params):
+    return _run_report(env, 'account_reports.cash_flow_statement', params)
+
+
+def compare_periods(env, params):
+    report_ref = params.get('report_ref', 'account_reports.profit_and_loss')
+    report = _get_report(env, report_ref)
+    if not report:
+        return {'error': f'Report {report_ref} not found'}
+
+    period1 = _run_report(env, report_ref, {
+        'date_from': params.get('period1_from'),
+        'date_to': params.get('period1_to'),
+    })
+    period2 = _run_report(env, report_ref, {
+        'date_from': params.get('period2_from'),
+        'date_to': params.get('period2_to'),
+    })
+    return {'period_1': period1, 'period_2': period2}
+
+
+def answer_financial_question(env, params):
+    question = params.get('question', '')
+    sql_query = params.get('sql_query')
+    if sql_query:
+        return {'error': 'Direct SQL not permitted. Use report tools instead.'}
+    return {'status': 'info', 'message': f'Use specific report tools to answer: {question}'}
+
+
+def export_report(env, params):
+    report_ref = params.get('report_ref', 'account_reports.profit_and_loss')
+    fmt = params.get('format', 'pdf')
+    report = _get_report(env, report_ref)
+    if not report:
+        return {'error': f'Report {report_ref} not found'}
+    date_opts = {}
+    if params.get('date_from'):
+        date_opts['date_from'] = params['date_from']
+    if params.get('date_to'):
+        date_opts['date_to'] = params['date_to']
+    options = report.get_options({'date': date_opts} if date_opts else {})
+
+    try:
+        if fmt == 'xlsx':
+            result = report.dispatch_report_action(options, 'export_to_xlsx')
+        else:
+            result = report.dispatch_report_action(options, 'export_to_pdf')
+
+        if isinstance(result, dict) and result.get('file_content'):
+            return {
+                'file_name': result.get('file_name', f'report.{fmt}'),
+                'file_type': result.get('file_type', fmt),
+                'file_content_b64': base64.b64encode(result['file_content']).decode(),
+            }
+        return {
+            'status': 'generated',
+            'message': f'Report exported as {fmt}. Use the Odoo UI to download.',
+        }
+    except Exception as e:
+        return {'error': f'Export failed: {str(e)}'}
+
+
+def get_invoicing_summary(env, params):
+    """Get invoicing summary — total invoiced by month, by partner, or for a date range.
+    Supports: monthly breakdown for a year, current month totals, or filtered by partner."""
+    from datetime import date, timedelta
+    import calendar
+
+    year = int(params.get('year', date.today().year))
+    partner_name = params.get('partner_name')
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+
+    domain = [
+        ('move_type', '=', 'out_invoice'),
+        ('state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ]
+
+    if partner_name:
+        partner = env['res.partner'].search([('name', 'ilike', partner_name)], limit=1)
+        if partner:
+            domain.append(('partner_id', '=', partner.id))
+        else:
+            return {'error': f'Partner not found: {partner_name}'}
+
+    if date_from and date_to:
+        domain += [('date', '>=', date_from), ('date', '<=', date_to)]
+        invoices = env['account.move'].search(domain, order='date desc')
+        total = sum(inv.amount_total for inv in invoices)
+        return {
+            'period': f'{date_from} to {date_to}',
+            'count': len(invoices),
+            'total': total,
+            'invoices': [{
+                'id': inv.id, 'name': inv.name, 'partner': inv.partner_id.name,
+                'date': str(inv.date), 'amount': inv.amount_total,
+                'payment_state': inv.payment_state,
+            } for inv in invoices[:30]],
+        }
+
+    # Monthly breakdown for the year
+    months = []
+    grand_total = 0
+    for month in range(1, 13):
+        m_start = f'{year}-{month:02d}-01'
+        last_day = calendar.monthrange(year, month)[1]
+        m_end = f'{year}-{month:02d}-{last_day}'
+        m_domain = domain + [('date', '>=', m_start), ('date', '<=', m_end)]
+        invoices = env['account.move'].search(m_domain)
+        total = sum(inv.amount_total for inv in invoices)
+        grand_total += total
+        months.append({
+            'month': f'{year}-{month:02d}',
+            'month_name': calendar.month_name[month],
+            'count': len(invoices),
+            'total': round(total, 2),
+        })
+
+    return {
+        'year': year,
+        'grand_total': round(grand_total, 2),
+        'months': months,
+        'partner': partner_name or 'All',
+    }
+
+
+def get_billing_summary(env, params):
+    """Get billing (vendor bills) summary — total billed by month or date range."""
+    from datetime import date
+    import calendar
+
+    year = int(params.get('year', date.today().year))
+    partner_name = params.get('partner_name')
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+
+    domain = [
+        ('move_type', '=', 'in_invoice'),
+        ('state', '=', 'posted'),
+        ('company_id', '=', env.company.id),
+    ]
+
+    if partner_name:
+        partner = env['res.partner'].search([('name', 'ilike', partner_name)], limit=1)
+        if partner:
+            domain.append(('partner_id', '=', partner.id))
+        else:
+            return {'error': f'Partner not found: {partner_name}'}
+
+    if date_from and date_to:
+        domain += [('date', '>=', date_from), ('date', '<=', date_to)]
+        bills = env['account.move'].search(domain, order='date desc')
+        total = sum(b.amount_total for b in bills)
+        return {
+            'period': f'{date_from} to {date_to}',
+            'count': len(bills),
+            'total': total,
+            'bills': [{
+                'id': b.id, 'name': b.name, 'partner': b.partner_id.name,
+                'date': str(b.date), 'amount': b.amount_total,
+                'payment_state': b.payment_state,
+            } for b in bills[:30]],
+        }
+
+    # Monthly breakdown
+    months = []
+    grand_total = 0
+    for month in range(1, 13):
+        m_start = f'{year}-{month:02d}-01'
+        last_day = calendar.monthrange(year, month)[1]
+        m_end = f'{year}-{month:02d}-{last_day}'
+        m_domain = domain + [('date', '>=', m_start), ('date', '<=', m_end)]
+        bills = env['account.move'].search(m_domain)
+        total = sum(b.amount_total for b in bills)
+        grand_total += total
+        months.append({
+            'month': f'{year}-{month:02d}',
+            'month_name': calendar.month_name[month],
+            'count': len(bills),
+            'total': round(total, 2),
+        })
+
+    return {
+        'year': year,
+        'grand_total': round(grand_total, 2),
+        'months': months,
+        'partner': partner_name or 'All',
+    }
+
+
+def get_collections_summary(env, params):
+    """Get payment collections summary — how much was collected (received) in a period."""
+    date_from = params.get('date_from')
+    date_to = params.get('date_to')
+    if not date_from or not date_to:
+        from datetime import date
+        today = date.today()
+        date_from = date_from or f'{today.year}-{today.month:02d}-01'
+        date_to = date_to or str(today)
+
+    payments = env['account.payment'].search([
+        ('payment_type', '=', 'inbound'),
+        ('state', '=', 'posted'),
+        ('date', '>=', date_from),
+        ('date', '<=', date_to),
+        ('company_id', '=', env.company.id),
+    ], order='date desc')
+
+    total = sum(p.amount for p in payments)
+    by_partner = {}
+    for p in payments:
+        pname = p.partner_id.name if p.partner_id else 'Unknown'
+        by_partner.setdefault(pname, {'count': 0, 'total': 0})
+        by_partner[pname]['count'] += 1
+        by_partner[pname]['total'] += p.amount
+
+    top_partners = sorted(by_partner.items(), key=lambda x: -x[1]['total'])[:15]
+
+    return {
+        'period': f'{date_from} to {date_to}',
+        'total_collected': round(total, 2),
+        'payment_count': len(payments),
+        'by_partner': [{'partner': k, 'count': v['count'], 'total': round(v['total'], 2)} for k, v in top_partners],
+    }
+
+
+TOOLS = {
+    'get_profit_loss': get_profit_loss,
+    'get_balance_sheet': get_balance_sheet,
+    'get_trial_balance': get_trial_balance,
+    'get_cash_flow': get_cash_flow,
+    'compare_periods': compare_periods,
+    'answer_financial_question': answer_financial_question,
+    'export_report': export_report,
+    'get_invoicing_summary': get_invoicing_summary,
+    'get_billing_summary': get_billing_summary,
+    'get_collections_summary': get_collections_summary,
+}