Cheshire Cat AI
Production-ready AI agent microservice — plug in any LLM, extend with Python plugins and hooks.
About
The Cheshire Cat is a framework for building conversational AI agents as a production microservice. One Docker container gives you a chatbot with pluggable LLMs, embedders, and vector memory per instance, an admin panel, a REST API, and a plugin system built on hooks and LangChain-style tools — write a Python file with @tool and @hook decorators and the Cat picks it up. Core plugins cover conversation history, memory, scheduling (White Rabbit), multimodality, and analytics.
It is a community-driven, GPL-3.0-licensed project (3,000+ stars, maintained through mid-2026) with a real plugin ecosystem and an active Discord — closer in spirit to a self-hosted conversational platform than to a code-first framework. The trade-offs: GPL-3.0 is copyleft, so check it against your distribution plans; the community is small next to the big frameworks; and the docs assume you are comfortable with Docker and Python packaging.
What it can do
- Conversational agent as a Docker microservice with admin panel
- Pluggable LLMs, embedders, and vector memory per chatbot instance
- Plugin system: `@hook` event system + `@tool` function calling
- Core plugins: history, memory, scheduling, multimodality, analytics
- REST API, Vue chat widget, and RabbitMQ hooks for scale-out
- Community plugin registry and active Discord
Setup
- Install Docker, then run:
docker run --rm -it -p 1865:80 ghcr.io/cheshire-cat-ai/core:latest. - Open the admin panel and configure your LLM and embedder (any supported provider).
- Chat via the admin panel, the Vue widget, or the REST API at
localhost:1865/docs. - Extend it: write a plugin (hooks + tools in Python) and drop it in the plugins folder or install from the plugin registry.
- For persistence, run with docker compose and mounted volumes.
Prerequisites
Docker; an LLM provider API key (or a local model endpoint); Python familiarity for plugin development.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (metrics tier, metrics-v1). Evidence: 3,094 GitHub stars (GitHub API), GPL-3.0, pushed Jul 29 2026; the official docs corroborate the plugin/hook architecture and the Docker-first deployment; an active maintained fork corroborates ongoing community use. Strengths: genuinely production-shaped (admin panel, API, scheduling, analytics), real plugin ecosystem, honest scope. Docked for: (1) GPL-3.0 copyleft — check before distributing or embedding; (2) small community vs mainstream frameworks; (3) last push ~2.5 months ago, slower cadence than peers. License: GPL-3.0.
★ 3,094 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.0/10
- Maintenance · recency, releases, issue load
- 8.0/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · Adoption 7: starScore from 3,094 stars on cheshire-cat-ai/core
- · Maintenance 8: pushed 2026-07-29 (recency 8); slower cadence than peers
- · Openness 10: GPL-3.0 is OSI-approved (0.7*10) + official docs verified
- · Manual 7: official docs corroborate the plugin/hook architecture and Docker-first deployment; an active fork corroborates community use; docked for GPL-3.0 copyleft, small community, slower release cadence
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.