#102 · Primary category: AI Gateway & API Infrastructure
harnessrouter
Self-hosted, Apache-2.0 unified API to run multiple agent harnesses (Codex, Claude Code, etc.) with sessions, streaming, and failure handling.
Project last updated:08/30/26
GitHub Stars
643
Forks
62
Contributors
9
License
Apache-2.0
Why we included this project
Running more than one coding agent usually means maintaining separate integrations for each one. HarnessRouter sits in front of Codex, Claude Code, Hermes, Pi, and DeepSeek's harness and exposes them through a single API, so swapping agents doesn't require touching the code that calls them. It also handles the unglamorous parts of agent orchestration, including sessions, streaming output, file access, cancellation, and failure handling, all inside one Docker container on your own hardware with your own provider keys. Since it implements the Unified Harness Protocol, an open standard, the same interface works across the harnesses it supports, which makes it a reasonable foundation for building tooling that talks to several agents instead of committing to one vendor's API.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
OmniRoute
Open-source AI gateway: one endpoint, 290+ providers (90+ free), 500+ models, auto-fallback, token compression, MCP/A2A, works with Claude Code, Codex, Cursor, Cline, Copilot.
CLIProxyAPI
Wrap Antigravity, ChatGPT Codex, Claude Code, Grok Build as an OpenAI/Gemini/Claude/Codex compatible API service, allowing you to enjoy the free Gemini 3.1 Pro, GPT 5.6 Series, Grok 4.5, Claude model through API
kong
🦍 The API and AI Gateway
novu
The open-source communication infrastructure for agents and products
GPT_API_free
Free API for large models, supports GPT, DeepSeek, etc., 10k free points daily, paid plans at 10-20% of official price.