china-qijizhifeng/agentic-harness-engineering
A research framework that runs a coding agent through repeated rounds: it reads the execution traces, distils them into reports, then proposes evidence-backed edits to the agent's prompts, tools and skills, and checks them in the next round.
Not tested yetKey files read, not run
How we know
We read the README and key files, not every source file, and did not run it, so the most we can say is Idea adapted or Risky.
Good for
Researchers and teams who want to tune a coding agent's setup from measured failures instead of guessing.
Not for
Everyday users; it needs sandboxed runs, several model keys and a benchmark-style loop to be useful.
Repo
- Stars
- 875
- License
- MITUse it, change it, sell it, closed-source included. Keep the author's copyright notice.
- Maturity
- active
- Last push
- 2026-08-03
- Last checked
- 2026-11-14