awesome-harness-engineering — the knowledge in this project is far more valuable than the number suggests — it contains frontline engineering practices from OpenAI, Anthropic, Microsoft, and Meta. GitHub: https://github.com/ai-boost/awesome-harness-engineering… What determines whether an agent works well is not the model, but the harness wrapped around the model. How context is fed, how tool interfaces are designed, how verification loops are run, how memory systems are managed — all of this together is called harness engineering. The materials in this project let you learn real-world engineering practices. LangChain took a coding agent from 30th place on Terminal Bench to the top 5 by changing the harness design. Microsoft’s Azure SRE agent autonomously handled over 35,000 production incidents — the postmortem is included here. Anthropic’s long-horizon task harness design guide, context engineering guide, and the reverse-engineering analysis paper of Claude Code’s internal architecture are all included. Also included: Martin Fowler’s survey on harness engineering, OpenAI’s Codex agent loop breakdown, covering the full pipeline of planning and task decomposition, context compression, permission sandboxing, observability, and human-in-the-loop.