What to Install

This module is typically auto-installed as a dependency of llm_assistant or llm_thread.

For AI with function calling:

odoo-bin -d your_db -i llm_assistant,llm_openai

Auto-Installed Dependencies

These are pulled in automatically:

• llm (core infrastructure)

Optional Enhancements

Common Setups

Using the @llm_tool Decorator (Recommended)

Zero boilerplate - just decorate your method and it's automatically available to LLMs.

from odoo import models
from odoo.addons.llm_tool.decorators import llm_tool

class ResUsers(models.Model):
    _inherit = "res.users"

    @llm_tool(read_only_hint=True, idempotent_hint=True)
    def get_system_info(self) -> dict:
        """Get basic Odoo system information"""
        return {
            "database_name": self.env.cr.dbname,
            "company_name": self.env.company.name,
            "user_count": self.env["res.users"].search_count([]),
        }

That's it! The tool is automatically:

• Registered in the database when Odoo starts
• Available to all LLM providers (Claude, ChatGPT, etc.)
• Description extracted from docstring
• Schema generated from type hints
• Validated with Pydantic

Decorator Options

@llm_tool(
    schema={...},              # Optional: Manual JSON schema
    read_only_hint=True,       # Tool only reads data
    idempotent_hint=True,      # Multiple calls have same effect
    destructive_hint=False,    # Tool modifies/deletes data
    open_world_hint=False,     # Tool interacts with external systems
)
def your_tool_method(self, param1: str, param2: int = 10) -> dict:
    """Tool description - shown to the LLM"""
    pass

Using Custom Implementation

For tools managed via XML data files, extend llm.tool:

class LLMTool(models.Model):
    _inherit = "llm.tool"

    @api.model
    def _get_available_implementations(self):
        implementations = super()._get_available_implementations()
        implementations.append(("my_custom_tool", "My Custom Tool"))
        return implementations

    def my_custom_tool_execute(self, param1, param2=None):
        """Execute your custom tool logic"""
        return {"result": "success"}

Then create tool records in XML:

<record id="my_custom_tool" model="llm.tool">
    <field name="name">my_custom_tool</field>
    <field name="implementation">my_custom_tool</field>
    <field name="description">Tool description for the LLM</field>
    <field name="input_schema">{"type": "object", "properties": {...}}</field>
</record>

Built-in Implementations

• odoo_record_retriever - Search and retrieve Odoo records
• odoo_record_creator - Create new records
• odoo_record_updater - Update existing records
• odoo_record_unlinker - Delete records
• odoo_model_method_executor - Execute any model method
• odoo_model_inspector - Inspect model structure and fields

Module Information

• Name: LLM Tool
• Version: 18.0.3.0.0
• Category: Technical
• License: LGPL-3
• Dependencies: llm
• Author: Apexive Solutions LLC

Key Models

• ``llm.tool``: Tool definition and execution
• ``llm.tool.consent.config``: User consent configuration