release-gate
Skill0 starsChanged 43 days agoMITView on GitHub
Run an exact-candidate release gate (freeze one source revision, run the declared checks once each, keep hashed evidence, get an independent evidence review, and only then call it ready). Use when asked whether something is ready to release or ship, to freeze a candidate, or to set up or run a release gate.
10 other projects use a version of this.
Adopted by
Most recently taken up by
- philoserf/claude-code-configtheir own version, first seen Nov 7, 2026
- kaandgnsh/Claude-Bluetheir own version, first seen Nov 1, 2026
- my-chiefmind/ai-maestrotheir own version, first seen Oct 13, 2026
- jfkconstruct/claude-skillstheir own version, first seen Oct 11, 2026
- imchangchang/dsh-onetheir own version, first seen Oct 11, 2026
- nickdesi/FFBB-MCP-Servertheir own version, first seen Sep 23, 2026
Other versions10 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 10
- philoserf/claude-code-config★ 14same name
- my-chiefmind/ai-maestro★ 6same name
- nickdesi/FFBB-MCP-Server★ 4same name
- wakqasahmed/ai-engineering-workflow-skills★ 2same name
- wakqasahmed/skills★ 2same name
- CaduS0uza/qa-supreme★ 1same name
- imchangchang/dsh-one★ 1same name
- jfkconstruct/claude-skills★ 0same name
- kaandgnsh/Claude-Blue★ 0same name
- youraioffice/your-ai-office-tools★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository6 other files it gives its agents.
- templates/AGENTS.mdAGENTS.md
- failure-to-ruleSkill
- handoffSkill
- refute-reviewSkill
- risk-tiersSkill
- sprint-specSkill
The file
Files
Exact-candidate release gate
Readiness belongs to one exact candidate: a source revision plus the artifacts built from it. "Implemented", "tested", "deployed", "enabled" and "healthy" are separate claims. The gate produces evidence. It publishes nothing, and a local pass is not permission to publish.
Template (declaration schema, step table, state table): ${CLAUDE_PLUGIN_ROOT}/templates/GATE.md. Command: /agent-operating-kit:gate.
Asked a general question with no candidate in scope: answer from the sequence and the blocker list below. Do not go looking for a repository to gate. Use the command when there is one.
Sequence
- Declare before coding. List every changed product path and, per outcome, the named executable checks per platform, with a written reason for each exclusion. The runner refuses a change whose files are not covered.
- Freeze. Record revision, dirty flag and content hash. Refuse a dirty tree. Take a lease on shared ports, data stores and devices. Refuse to start rather than share.
- Run the steps in order, each passing exactly once. Gate self-test first: every rule must fire on its failing fixture and stay quiet on its passing one.
- Keep evidence. One private directory per run with the report, raw logs, and every artifact with its SHA-256. Keep blocked runs. Never overwrite or relabel them.
- Re-check stability. Revision and hash unchanged at the end, or the run is void.
- Review independently, in fresh contexts. Partition the evidence with no gaps or overlaps, recompute every hash, grade what matters, and report evidence quality (byte-identical captures under different names, mislabeled files).
- Finalize separately. A step that re-derives everything from files decides
ready, and only for the platforms covered.
What blocks
Skips, retries, focused tests, expected failures, flaky passes, zero tests found, stale artifacts, a source change mid-run, capture-only or partial-platform runs, and a baseline updated to clear a failure.
States
blocked, checks-passed, review-pending, partial, ready. The runner can reach checks-passed at most. It never declares ready itself.
Repair loop
Classify the failure. Reproduce a product defect with an assertion before fixing it. Never weaken an assertion. A wait can delay a failure but not convert one. A failure that appears only when a phase runs alone is a defect until disproven. Stop after three attempts on the same failure and report. New source is a new candidate and a new run.
State the limits
List what the gate does not certify: other platforms, signing, the production network, deployment, live health.
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.

