Harness Repair
Repair failing lint, typecheck, test, build, CI, migration dry-run, or manual acceptance loops without hiding failures. Binds failures to the CodeRail Coordinate V and P fields and writes trace events. Use when the done gate fails.
Install
$ agentstack add skill-haipingshi-coderail-harness-repair ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Links to your public security report.[](https://agentstack.voostack.com/security/report/skill-haipingshi-coderail-harness-repair) Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Harness Repair
Treat failing harness as signal, not noise. Failures belong to the coordinate's V field; reusable lessons belong to P → LESSONS; the repair result belongs to V and to the trace log.
Procedure
- Identify the exact failing command (this is a V entry).
- Capture a short failure summary, not a full log.
- Classify failure:
- implementation bug
- wrong test
- stale fixture
- environment issue
- missing dependency
- task contract mismatch
- North-Star mismatch
- Fix the smallest root cause.
- Re-run the focused command (V).
- Re-run global checks if the fix touches shared code.
- Update
docs/LESSONS.mdonly if the lesson is reusable (P → LESSONS). - Write trace events:
verifywithharness_result: failedfor the original failure (command + summary + evidence location only).changefor the fix (files modified).verifywithharness_result: passedafter the re-run.lessononly if the failure is reusable.
- If failure repeats twice with unclear root cause, stop and report (X trigger).
Rules
- Do not weaken tests just to pass.
- Do not skip failing commands.
- Do not mark done with failed harness.
- Do not broaden scope (S) without updating the task contract.
- Do not record full test output in trace — only the command, result, summary,
and where the evidence lives.
- If the harness itself is wrong, create or update a harness task and record
the decision.
Output
## Harness Repair Report
Failing command (V):
Failure summary:
Failure class:
Root cause:
Fix applied (files in S):
Re-run result (V):
Global checks:
Lesson needed (P → LESSONS): yes | no
Trace events: verify(failed) | change | verify(passed) | lesson
Task can proceed to done: yes | noSource & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: HaipingShi
- Source: HaipingShi/coderail
- License: MIT
- Homepage: https://github.com/HaipingShi/coderail
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.