# Github

> Github is an open-source AI tool for developer tools. Lightweight async Python client for multiple LLM providers with agent and MCP support. It is available on API, CLI and self-hosted.

- **Kind:** Tool
- **Pricing:** Open source
- **Platforms:** API, CLI, self-hosted
- **Website:** https://polarsen-io.github.io/padwan-ai/
- **Categories:** [developer tools](https://ai-powered.com/categories/developer-tools)
- **Listing updated:** 2026-11-18

## About

The README says Padwan AI is a Python library that talks to several model providers directly over their HTTP APIs instead of wrapping each vendor's SDK. Its only runtime dependency is niquests, and request and response shapes are plain TypedDicts, so there is no validation layer to add at runtime.

On top of the provider clients, it offers an AgentSession that loops between a model and tool calls until the model returns a plain text answer. Tools can come from individual MCP tool objects or from whole Streamable HTTP or stdio MCP servers. Realtime voice sessions, Gemini and ElevenLabs text-to-speech, and thought-token streaming for thinking models are also covered, though several provider integrations are marked experimental.

The project suits Python developers who want provider flexibility without adopting a large framework. A separate padwan-cli package provides a terminal interface on top of the library.

## Features

- **Unified multi-provider interface:** One client API covers OpenAI, Gemini, Mistral, Grok, Anthropic and any OpenAI-compatible endpoint.
- **Async agentic loop:** AgentSession runs multi-turn conversations with tool dispatch, parallel execution and approval hooks.
- **MCP server connections:** Connects to Model Context Protocol servers over streamable HTTP or stdio transports.
- **Streaming and thinking tokens:** Streams output for all providers and routes thought tokens separately via a callback.
- **Realtime voice sessions:** Speech-to-speech WebSocket sessions for OpenAI Realtime, Gemini Live and Grok Voice.
- **Typed, minimal dependencies:** Type hints throughout with TypedDicts and niquests as the only listed dependency.

## Good for

- Calling several LLM providers from one Python codebase without installing each vendor SDK
- Building a tool-using agent that reads tools from MCP servers
- Streaming model reasoning tokens separately from final answers
- Adding speech-to-speech or text-to-speech output to a Python application

## FAQ

### Does it require provider SDKs?

No. The README says Padwan talks to each API directly and has no provider SDK dependencies.

### How do I install it?

Run pip install padwan-ai, or uv add padwan-ai when using uv.

### Which MCP transports are supported?

Streamable HTTP and stdio, through the McpStreamable and McpStdio classes.

### Is there a command-line interface?

The README points to a separate package, padwan-cli, for the interactive CLI and TUI.

### Is Github free?

Yes. Github is open source, so you can use and self-host it for free.

### What are the best alternatives to Github?

The closest alternatives to Github are ruby_llm, cc-switch, gateway, Lynx and archestra.

## Where this comes from

- Description: written from the project's website.

## Alternatives

- [ruby_llm](https://ai-powered.com/tools/ruby-llm): ruby_llm is an open-source AI tool for developer tools.
- [cc-switch](https://ai-powered.com/tools/cc-switch): cc-switch is an open-source AI tool for developer tools.
- [gateway](https://ai-powered.com/tools/gateway-10): gateway is an open-source AI tool for developer tools.
- [Lynx](https://ai-powered.com/tools/lynx-3): Lynx is an open-source AI tool for developer tools.
- [archestra](https://ai-powered.com/tools/archestra): archestra is an open-source AI tool for security.
- [Aiproxy](https://ai-powered.com/tools/aiproxy): Aiproxy is an open-source AI tool for developer tools.
- [mcp-use](https://ai-powered.com/tools/mcp-use): mcp-use is a freemium AI tool for developer tools.
- [bifrost](https://ai-powered.com/tools/bifrost-3): bifrost is an open-source AI tool for ai models and llm tooling.

---
Source: https://ai-powered.com/tools/github-9
