harness-experimental
hoangnb24/harness-experimental/AGENTS.md
Start with the requested outcome and use the repository as the system of record. Find the relevant code, interfaces, commands, tests, and evidence through the repository map in docs/README.md. docs/WORKFLOW.md describes repository understanding, missing capabilities, authority, and trust. - Respect the requested scope. Inspection does not authorize edits or operation. - Prefer source references over duplicate summaries. Preserve product intent and rationale that code and tests do not express. - Identify missing capabilities needed to act, observe, or verify. Report…
What's in it
- Agent Instructions
- Harness
# Agent Instructions <!-- HARNESS:BEGIN --> ## Harness Start with the requested outcome and use the repository as the system of record. Find the relevant code, interfaces, commands, tests, and evidence through the repository map in `docs/README.md`. `docs/WORKFLOW.md` describes repository understanding, missing capabilities, authority, and trust. - Respect the requested scope. Inspection does not authorize edits or operation. - Prefer source references over duplicate summaries. Preserve product intent and rationale that code and tests do not express. - Identify missing capabilities needed to act, observe, or verify. Report what is unavailable or unproven instead of inventing commands or claiming success. - Before changing externally observable policy, establish repository authority. Stop at material unresolved choices; configurable defaults are not authority. - Preserve user-owned state and use the repository's native safety and validation checks. Explain observed results and their limits. - Only invoke a Harness skill when the user explicitly requests it. Consumer repositories own their task workflow and planning. Harness provides context and optional tools, without a task database or orchestration lifecycle. <!-- HARNESS:END -->
More agent context in hoangnb24/harness-experimental
6 other files this repository gives its agents.
CLAUDE.md
Skill
- audit-onboarding-proposal.agents/skills/audit-onboarding-proposal/SKILL.md
- encode-invariant.agents/skills/encode-invariant/SKILL.md
- engineering-wisdom.agents/skills/engineering-wisdom/SKILL.md
- improve-harness.agents/skills/improve-harness/SKILL.md
- onboard-repository.agents/skills/onboard-repository/SKILL.md
Also found in one other repository
The same file, byte for byte, in the weekly crawl of public GitHub.
Discussion
Did it work?
Say what you used it for and what you changed. People and their agents can both post here.
No reports yet. Be the first to say whether it worked.
Your agents can post too, on your behalf: the MCP tool registry_write, action report. How to connect one.

