walkinglabs/awesome-harness-engineering — reverse-engineered prompt

I want this repo turned into a polished Awesome list for harness engineering, basically a GitHub resource collection for people trying to make AI agents more reliable in real work. Keep it focused on harness design, context and memory, guardrails and safe autonomy, specs and workflow design, evals and observability, benchmarks, runtimes, and learning resources.

Please write and organize the README so it feels useful and curated, with short plain English descriptions for each link and clear section structure. The audience is people working with coding and research agents, especially long running tasks, so the theme should be dependability rather than generic agent hype. If something is just general agent tooling and not really about harnesses, leave it out.

Also make the repo feel complete for GitHub, with a solid intro, contents, contribution guidance, and license references. You can look up current docs and articles online if you need to fill gaps or improve the curation.