Custom Tools
Custom Tools let you extend the power of CMND.ai by connecting LLMs to your own logic, services, and infrastructure.
They allow you to expose internal functions — whether it’s querying a database, triggering backend workflows, or calling private APIs — and make them accessible within natural conversations powered by LLMs.
This means your assistants can do more than talk — they can act.
Ways to Create Custom Tools
You can build and connect custom tools in CMND.ai in two ways:
-
CMND Extensions API: Develop your tool externally using your own infrastructure (e.g., FastAPI, Flask, Node.js), then connect it to CMND via a secure public endpoint.
-
CMND Extensions Studio: Create and deploy your tool entirely within CMND’s web platform — no separate hosting needed. Ideal for quick setups, prototypes, or simpler tools.
Both options give you full flexibility over your tool’s behavior, schema, and integration.
What You Can Do with Custom Tools
-
Connect LLMs to your backend logic
Wrap any Python or JavaScript function and expose it as a callable tool inside CMND. -
Build task-specific utilities
From checking inventory to performing calculations or generating content, tools can be scoped to exactly what your assistant needs. -
Access private or dynamic data
Pull real-time information from your internal systems that isn’t available to the LLM by default. -
Maintain control over execution
Tools run in your environment, so you can manage permissions, validation, and responses as needed. -
Make assistants actionable
Go beyond chat — enable your assistant to trigger workflows, run queries, or fetch structured responses in real-time.
Example Use Cases
- A customer support assistant that queries recent orders via a backend API
- A product advisor that fetches real-time pricing from your database
- A data assistant that performs filtered SQL lookups based on user prompts
- A creative tool that wraps your own prompt engineering logic and templates
Custom Tools are a powerful way to bring domain-specific intelligence and capability into your CMND-powered assistant — securely, flexibly, and in the language of your own codebase.