agentleFS
sabirmgdsabirmgd/terminal-first-goal-loop/skills/release-proof/SKILL.md

release-proof

Skill2 starsChanged 2 months agoMITView on GitHub

Prove that a candidate is truly ready or live by binding source, CI, artifact, runtime, and consumer evidence to the same exact version. Use when a feature, fix, preview, sandbox, or release should not be trusted from a green pipeline alone.

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
  1. Distinctive:

    Use this skill whenever a green pipeline is not enough.

    line 8
  2. Distinctive:

    Every verdict must bind to one exact candidate identity such as a commit SHA, image digest, artifact version, or deployment revision.

    line 12
  3. Very specific:

    If any layer is missing or mismatched, the verdict is not pass.

    line 23
  4. Distinctive:

    Run the real browser, API, CLI, or consumer journey.

    line 41
  5. Distinctive:

    Do not judge a rollout while it is still converging.

    line 44
  6. Very specific:

    Stop with pass only when all required layers point to the same candidate and the consumer proof passes.

    line 68

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Release Proof

Use this skill whenever a green pipeline is not enough.

One Candidate, Five Layers

Every verdict must bind to one exact candidate identity such as a commit SHA, image digest, artifact version, or deployment revision.

Prove these layers in order:

  1. source
  2. ci
  3. artifact
  4. runtime
  5. consumer

If any layer is missing or mismatched, the verdict is not pass.

Rules

  1. Runtime proof must name the exact environment.
  2. For UI claims, browser proof is mandatory.
  3. For API-only claims, use the real client path or exact request shape.
  4. A healthy runtime is not enough if it is serving the wrong candidate.
  5. A green CI run is not enough if runtime or consumer proof is missing.
  6. This skill proves state only. It does not authorize merge, deploy, retry, or rollback.

Workflow

  1. Freeze the candidate identity and the environment being judged.
  2. Prove the source branch or commit is the intended candidate.
  3. Collect the required CI runs for that exact candidate.
  4. Prove the built image, package, or artifact was produced from it.
  5. Prove the target runtime is ready and serving that artifact.
  6. Run the real browser, API, CLI, or consumer journey.
  7. Record every missing or mismatched layer as blocked or fail.

Do not judge a rollout while it is still converging. Wait for the terminal state, then re-read the live artifact and traffic identity.

Common False Greens

  • CI passed for one commit, but another commit was deployed.
  • The expected artifact exists, but traffic still points to the old revision.
  • The service is healthy, but required migrations or runtime configuration are missing.
  • The backend endpoint works, but the browser journey is broken.
  • A deployment workflow succeeded, but authenticated smoke tests were skipped.

Output Contract

RELEASE PROOF | candidate=<id> | verdict=<pass|blocked|fail>
SOURCE | <proof>
CI | <proof and result>
ARTIFACT | <proof and result>
RUNTIME | env=<name> | <proof and result>
CONSUMER | <browser, api, or client proof>
NEXT | <none or exact remaining gate>

Stop Conditions

Stop with pass only when all required layers point to the same candidate and the consumer proof passes. Stop with blocked or fail as soon as an unresolved layer prevents a truthful verdict.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.