#8888th of 107 in Coding agents and AI editors
tau
Python terminal coding agent with a small, layered, readable codebase
Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 3k
- License
- MIT
- Last commit
- Oct 2026
- Last release
- Oct 2026
- Language
- Python
1 on 9 Jun 20262,977 on 20 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Tau is a terminal coding agent that reads files, edits code and runs shell commands from a Textual TUI or a one-shot print mode. Sessions are stored as append-only JSONL under ~/.tau/sessions/. The code is split into three packages (tau_ai, tau_agent, tau_coding), so the agent harness can be reused as a library. It is also written as a teaching example of how coding agents are built.
Who it is for: Developers who want a terminal coding agent or a codebase to learn from
Strengths
- Three-layer split lets the tau_agent harness run without the CLI or TUI
- Durable JSONL sessions with resume, branching and compaction
- Supports OpenAI, Anthropic, OpenRouter, Hugging Face and OpenAI-compatible endpoints
- Reads AGENTS.md, .tau/ and .agents/ resources; supports skills and prompt templates
Weaknesses
- Only four built-in tools: read, write, edit and bash
- README says it is under active development
- Requires Python 3.12 or newer
- README does not mention MCP support
What it needs
- no GPU
- Needs Python 3.12+, uv (installed by the install script), model provider account or endpoint
- Models: OpenAI, Anthropic, OpenAI Codex subscription, OpenRouter, Hugging Face
Also in Coding agents and AI editors
See all 107| Rank | Project | Score |
|---|---|---|
| 1 | claw-codeRust CLI coding agent harness using Anthropic and OpenAI-compatible APIs | 83 out of 100 |
| 2 | ArchonYAML workflow engine that runs AI coding agents through fixed dev steps | 83 out of 100 |
| 3 | Open SWEAgent that turns issues into pull requests, reviews PRs and monitors CI | 81 out of 100 |
| #4 | oh-my-openagentTerminal coding agent with multi-model orchestration and git-backed memory | 80 out of 100 |
| #5 | gooseRust AI agent with desktop app, CLI and API, extended through MCP | 80 out of 100 |