merge-gate-auditor
Skill1 starsChanged 2 months agoApache-2.0View on GitHub
Sample unlabelled PR records from merge-gate-data, prepare structured Audit Verdict Cards, and stage human verdicts. Use whenever sampling unlabelled audit candidates, conducting merge-gate audit triage, or reviewing pull-request outcomes against the calibrated merge gate.
In this repository39 other files it gives its agents.
- agent-core/AGENTS.mdAGENTS.md
- .agents/AGENTS.mdAGENTS.md
- AGENTS.mdAGENTS.md
- behavioral-regression/AGENTS.mdAGENTS.md
- claude-foundation/AGENTS.mdAGENTS.md
- config/AGENTS.mdAGENTS.md
- corpora/AGENTS.mdAGENTS.md
- demo/AGENTS.mdAGENTS.md
- docs/AGENTS.mdAGENTS.md
- examples/AGENTS.mdAGENTS.md
- experiments/AGENTS.mdAGENTS.md
- flow-corpus/AGENTS.mdAGENTS.md
- flow-protocol/AGENTS.mdAGENTS.md
- openspec/AGENTS.mdAGENTS.md
- scripts/AGENTS.mdAGENTS.md
- skills/AGENTS.mdAGENTS.md
- tests/AGENTS.mdAGENTS.md
- claude-foundation/CLAUDE.mdCLAUDE.md
- update-executive-reportSkill
- update-executive-reportSkill
- architecture-drift-guardSkill
- commonSkill
- corpus-guardianSkill
- dataset-lintSkill
- deploySkill
- e2e-matrix-sentinelSkill
- eval-corpus-forgeSkill
- hierarchical-recursive-brainstormSkill
- model-benchSkill
- openai-judgeSkill
- openspec-implementation-reviewSkill
- openspec-peer-reviewSkill
- openspec-quality-planSkill
- pre-pr-gateSkill
- project-setupSkill
- quality-gateSkill
- refactoring-decomposerSkill
- repo-invariant-reviewSkill
- test-completeness-guardSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 4 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 12
Zero Synthetic Audits: Never synthesize or fabricate HUMAN_AUDIT records.
- Very specific: line 14
Plumbing Store Sync: All reads and writes to merge-gate-data operate through agent_core.store_sync plumbing (mktree, commit-tree), never …
- Distinctive: line 44
Verify if any post-merge revert or fix-up was committed within the 7-day observation window.
- Distinctive: line 45
Check behavioral regression test status:
- Distinctive: line 62
After human maintainer executes verdicts:
- Distinctive: line 71
No synthetic records are generated.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Merge Gate Auditor — Human Audit Triage Skill
The merge-gate-auditor automates the sampling of unlabelled pull request records from origin/merge-gate-data, formats detailed diff and behavioral contexts, and constructs Audit Verdict Cards for review by the human merge owner (@ianshank).
Core Invariants
- Zero Synthetic Audits: Never synthesize or fabricate
HUMAN_AUDITrecords. The calibrated auto-merge gate requires genuine human audit evidence ($N \ge 200, ECE \le 0.05, AUROC \ge 0.65$). - Cold-Start Floor: The audit sampler enforces
AuditConfig.per_domain_floor(default 30, workflow override 3) before a domain can exit cold-start. - Plumbing Store Sync: All reads and writes to
merge-gate-dataoperate throughagent_core.store_syncplumbing (mktree,commit-tree), never touching working trees or triggering CI ([skip ci]).
Procedure
1. Pull Latest Store Records
python -m agent_core.store_sync pull --repo-dir . --store data/merge_outcomes.jsonl2. Inspect Soak Progress
python -m agent_core.store_sync stats --store data/merge_outcomes.jsonl --soak-progress --audit-floor 3Verify:
- Total records in store ($N$)
- Number of human audits accumulated per domain
- Cold-start status per domain
- Shortfall vs soak target
3. Sample Unlabelled Candidates
python -m agent_core.audit_sampler --store data/merge_outcomes.jsonl select --per-domain-floor 3 --with-propensity4. Generate Audit Verdict Card
For each sampled candidate change ID:
- Inspect git log and diff of the change:
git show <change_id> - Verify if any post-merge revert or fix-up was committed within the 7-day observation window.
- Check behavioral regression test status:
python -m pytest behavioral-regression/tests/ - Render the Audit Verdict Card:
### Audit Verdict Card: `<change_id>` - **Domain**: `<domain>` - **Merged At**: `<timestamp>` - **Merge Gate Confidence**: `<raw_confidence>` - **Revert / Fixup Observed**: None / Commit `<sha>` - **Maintainer Actions**: - Confirm Correct: `python scripts/record_audit_verdict.py --store data/merge_outcomes.jsonl --change-id <change_id> --actor @ianshank --correct` - Confirm Regressed: `python scripts/record_audit_verdict.py --store data/merge_outcomes.jsonl --change-id <change_id> --actor @ianshank --incorrect`
5. Sync Human Verdicts to Remote Store
After human maintainer executes verdicts:
python -m agent_core.store_sync push --repo-dir . --store data/merge_outcomes.jsonl --actor @ianshankDefinition of Done
- Unlabelled candidates are sampled strictly according to the domain cold-start floor.
- Verdict cards provide complete commit, diff, and revert context.
- No synthetic records are generated.
- Updated store records pass
agent_core.store_sync statsvalidation.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

