eliasforge/fable-agent-orchestration/contributor-evidence-gate

Claude Code and agent skill by eliasforge/fable-agent-orchestration

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Keep public repository contributions narrow, current, evidence-backed, and easy to review. Use before editing a public repo, preparing an AI-assisted patch, or writing a pull request that needs a clean task contract, live-state check, scope fence, adversarial check, and concise report.

Install

npx vskill@latest install eliasforge/fable-agent-orchestration/contributor-evidence-gate

SKILL.md

Contributor Evidence Gate

Use this when a public repository contribution must be small enough to review and concrete enough to prove. The goal is not to make a patch look disciplined. The goal is to prevent broad, stale, or unverifiable work from entering the queue.

When To Use

Use for:

  • small public issues;
  • AI-assisted pull requests;
  • documentation changes that mention current behavior;
  • test or CI changes;
  • contributor task packs;
  • external patches that need a clear review contract.

Do not use as a substitute for the repository's actual tests, CI, security review, or release gates.

The Loop

  1. Check current truth.
  • Read the current files, branch, open issue, and relevant tests.
  • Do not rely on stale README text when code or CI can answer the question.
  1. Write the patch contract before editing.

``text Task: Files expected: Non-scope: Acceptance command: Risk if wrong: ``

  1. Fence the scope.
  • Change only the files needed for the task.
  • Put nearby issues in Noticed, not changed.
  • Do not silently refactor adjacent areas.
  1. Build one narrow change.
  • One behavior, contract, or documentation correction per pull request.
  • If the patch needs two unrelated proofs, split it.
  1. Prove the task, not the vibe.
  • Run the command that would catch the actual failure.
  • A generic green check is not enough if it does not exercise the changed

behavior.

  1. Run an adversarial pass.
  • Name one realistic way the patch could still be wrong.
  • Check that risk against the diff, test, generated artifact, or live output.
  1. Keep public memory clean.
  • Do not commit secrets, private notes, raw prompts, raw model outputs,

bulky logs, or local-only paths unless the repo explicitly asks for them.

  • Persist durable facts, not chat residue.
  1. Report compactly.
  • Keep the final pull request summary short.
  • Lead with what changed and how it was proved.

Pull Request Shape

text
Task:
Files changed:
Non-scope:
Proof command:
Proof result:
Adversarial check:
Noticed, not changed:

Review Gates

Ask these before submitting:

  • Would this patch still look correct if the README were stale?
  • Would the proof fail if the changed behavior were broken?
  • Did the patch change only what the task required?
  • Did it avoid raw secrets, raw prompts, raw outputs, and local-only private

material?

  • Can a reviewer reproduce the claim from the repo and the listed command?

Failure Modes

Reject or split the work when:

  • the patch combines feature work, refactoring, generated artifacts, and docs;
  • the proof is only "tests pass" but the task is more specific;
  • the PR summary hides residual risk;
  • the patch makes a capacity, security, or reliability claim without a measured

gate;

  • the contribution imports external workflow text as authority instead of

rewriting it for this repo's public rules.

Output

Return:

text
Patch contract:
Changed files:
Proof:
Adversarial check:
Residuals:
Next smallest task:

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/eliasforge/fable-agent-orchestration/contributor-evidence-gate/badge)](https://verified-skill.com/skills/eliasforge/fable-agent-orchestration/contributor-evidence-gate)

Something wrong with this skill or its page? Tell us.