eliasforge/fable-agent-orchestration/contributor-evidence-gate
Claude Code and agent skill by eliasforge/fable-agent-orchestration
- community
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-gateSKILL.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
- 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.
- Write the patch contract before editing.
``text Task: Files expected: Non-scope: Acceptance command: Risk if wrong: ``
- Fence the scope.
- Change only the files needed for the task.
- Put nearby issues in
Noticed, not changed. - Do not silently refactor adjacent areas.
- Build one narrow change.
- One behavior, contract, or documentation correction per pull request.
- If the patch needs two unrelated proofs, split it.
- 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.
- 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.
- 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.
- Report compactly.
- Keep the final pull request summary short.
- Lead with what changed and how it was proved.
Pull Request Shape
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:
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
- SKILL.md in the repositoryCheckingskills/contributor-evidence-gate/SKILL.md
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
[](https://verified-skill.com/skills/eliasforge/fable-agent-orchestration/contributor-evidence-gate)