Cheshire Cat AI
Cheshire Cat AI is a production-ready AI framework that enables the creation of AI agents using Python. It offers extensive functionality f...
Last verified:
What is Cheshire Cat AI?
Cheshire Cat AI is an open-source, production-ready framework that allows developers and AI builders to easily build and deploy custom AI agents. The framework is API-first and microservice-based, designed to add a conversational layer to pre-existing applications. Once installed and connected to a Large Language Model (LLM), it can be queried through REST APIs and WebSocket endpoints, returning responses from the connected LLM.
Key features include built-in RAG (Retrieval-Augmented Generation) with Qdrant for document-based knowledge, episodic memory for storing conversation history, and declarative memory for uploaded documents (PDF, TXT, Markdown, JSON, web pages). The framework supports tools for function calling (enabling the LLM to execute Python functions like calling REST APIs, querying databases, or managing domotics), hooks for event callbacks to customize agent behavior, and conversational forms for goal-oriented multi-turn conversations based on Pydantic models.
Cheshire Cat AI is designed for developers, AI engineers, and technical teams who want to create custom AI agents with extensibility via plugins. The framework is 100% dockerized with live reload, includes an admin panel at localhost:1865/admin for managing installations, installing plugins, visualizing memory, configuring LLMs and embedders, and managing users. It supports any language model via LangChain, offers multi-user support with granular permissions, and provides community-built client libraries in TypeScript/JavaScript, Python, C#, PHP, Ruby, Go, and Java, plus frontend widgets for Vue, Alpine, Svelte, React, and Angular.
Cheshire Cat AI pricing
Pricing model: Free
Cheshire Cat AI is completely free and open-source under the GNU General Public License v3.0 (GPL-3.0). There is no paid tier or subscription plan from the official project. Users only need to pay for their chosen LLM provider's API costs (such as OpenAI, or use free open-source LLMs via Ollama/vLLM). Paid support and customization services are available upon contact for those who need professional assistance.
Cheshire Cat AI pros
- Open-source with GPL-3.0 license
- Built-in RAG with Qdrant vector database
- 100% dockerized with live reload support
- API-first microservice architecture
- WebSocket chat with token streaming
- Extensible via plugins (just a folder)
- Admin panel for easy management
- Supports any LLM via LangChain
- Tools enable function calling for actions
- Hooks for event callbacks customization
- Conversational forms for multi-turn conversations
- Episodic memory for conversation history
- Declarative memory for document uploads
- Multi-user with granular permissions
- Community registry with one-click plugin install
- Client libraries in 7+ programming languages
- Frontend chat widgets for 6+ frameworks
- Integrations with Telegram, Discord, WordPress
- White Rabbit component for scheduled actions
- Active Discord community with 2.8k GitHub stars
Cheshire Cat AI cons
- No ready-to-use UI for end users (API-first only)
- Requires Docker and docker-compose installed
- First docker-compose run takes several minutes
- Docker images occupy several GBs
- No built-in paid hosting solution
- Plugins must be open-source and hosted on GitHub
- Manual UI implementation required for production
- Learning curve for hooks, tools, and forms
- Version 2 still in development (bug fixes only for v1)
- Requires API key from LLM provider separately
Frequently asked questions about Cheshire Cat AI
What is Cheshire Cat AI?
Cheshire Cat AI is an open-source, hackable, and production-ready framework that allows you to easily build and deploy custom AI agents. It is API-first and microservice-based, designed to add a conversational layer to pre-existing applications. Once connected to an LLM, it returns responses through REST APIs and WebSocket endpoints.
How do I install Cheshire Cat AI?
You need Docker installed on your system. The quickest way is: docker run --rm -it -p 1865:80 ghcr.io/cheshire-cat-ai/core:latest. For production, use docker-compose with the provided configuration. After running, chat at localhost:1865/admin and test the REST API at localhost:1865/docs.
What LLMs does Cheshire Cat support?
Cheshire Cat supports any language model via LangChain. You can use commercial LLMs like OpenAI's GPT models (gpt-3.5-turbo, gpt-4) by providing your API key, or open-source models via Ollama, vLLM, or other local runners. The framework is model-agnostic and lets you choose your preferred provider.
How do I upload documents for RAG?
You can upload PDF, TXT, Markdown, JSON files, and web pages through the Admin Portal at localhost:1865/admin or via the REST API. The Rabbit Hole component handles document ingestion, storing them in the declarative memory database (Qdrant). The Cat will then use this knowledge when answering questions.
What are plugins and how do I create one?
A plugin is just a folder in cat/plugins containing a Python file with hooks, tools, and forms. Create a folder in cat/plugins, add a Python file, and use decorators like @hook for event callbacks, @tool for function calling, or @form for conversational forms. Debug with live reload in the admin chat.
What are hooks in Cheshire Cat?
Hooks are event callbacks that let you customize your agent at predefined adaptation points in the main process flow. You can change the system prompt, inspect/edit inbound and outbound messages, fully customize the agent, or create pipelines for memory and uploads. Use the @hook decorator to attach Python functions.
What are tools and how do they work?
Tools enable LLM function calling by allowing the LLM to execute Python functions. You can manage domotics, call REST APIs, query databases, or integrate with symbolic reasoners. Use the @tool decorator on Python functions. The LLM automatically calls these tools when relevant based on the conversation context.
What are conversational forms?
Forms handle goal-oriented multi-turn conversations. Based on Pydantic models, they gather complex information on autopilot, support type validation and custom validation, ask for final confirmation (if ask_confirm=True), and have custom submit callback functions. Use the @form decorator with a CatForm class.
How do I integrate Cheshire Cat with my application?
Cheshire Cat is API-first, so you can integrate via REST API or WebSocket. Community client libraries exist for TypeScript/JavaScript, Python, C#, PHP, Ruby, Go, and Java. Frontend widgets are available for Vue, Alpine, Svelte, React, and Angular. You can also use Telegram bot, Discord bot, or WordPress plugin integrations.
Is there paid support available?
Yes, paid support and customization services are available for those who need professional assistance. The project is open-source under GPL-3.0, but if you need help with implementation, custom features, or enterprise support, you can contact the team for paid services.