Skip to content

Code of Conduct

1. Our Pledge

We want this repository to be a place where people can discuss skill design, evaluation, and examples rigorously without turning technical disagreement into personal conflict.

In all project spaces, we commit to: - communicate respectfully and directly - challenge ideas with evidence, not insults - keep safety, privacy, and responsible disclosure in mind - make room for newcomers, experts, and different working styles

2. Expected Behavior

  • Give concrete, evidence-based feedback in Issues, PRs, Reviews, and Discussions.
  • Critique methodology, skills, reports, and examples without attacking the contributor.
  • When disagreeing, explain the reasoning, assumptions, and trade-offs.
  • Flag unsafe content, secret exposure, or harmful guidance responsibly.
  • Keep bilingual governance/docs aligned when you touch them.
  • Prefer high-signal contributions over noise, flamebait, or content dumping.

3. Unacceptable Behavior

  • Harassment, insults, threats, discrimination, or personal attacks.
  • Repeated bad-faith accusations (for example, alleging fraud or dishonesty without evidence).
  • Deliberate disruption, trolling, brigading, or low-signal spam.
  • Posting secrets, private data, exploit payloads, or unsafe instructions irresponsibly.
  • Using AI-generated bulk content to flood discussions without review or accountability.
  • Any conduct that violates GitHub policies or applicable law.

4. Scope

This Code of Conduct applies to: - repository Issues, PRs, Reviews, Discussions, comments, and linked examples - public project-related conversations where someone is clearly acting as a project participant - governance, methodology, evaluation, and output-example discussions related to this repository

5. Reporting

If you experience or witness unacceptable behavior:

  1. For non-sensitive conduct issues, open an Issue with prefix [CoC] and include a factual timeline.
  2. For sensitive cases, request non-public follow-up and avoid posting private details publicly.
  3. If the issue also involves secrets, exploit details, or unsafe content, follow SECURITY.md instead of posting full details publicly.

Useful report details: - where it happened - links or screenshots if available - factual context - what outcome you want

6. Enforcement Principles

Maintainers will review reports based on available evidence and aim to provide an initial response within 3 business days.

Enforcement principles: - fairness: decisions should be evidence-based - privacy: share only what is necessary - proportionality: mild issues may start with warning/correction; severe or repeated issues may escalate - project fit: this repository allows strong technical disagreement, but not abusive behavior

7. Possible Consequences

Depending on severity, maintainers may: - issue a warning - ask for edits or removal of inappropriate content - lock or limit participation in a thread - temporarily suspend participation - permanently ban participation

8. Relationship to Security

This repository contains skills, examples, and helper scripts that may influence real-world AI behavior.

If a conduct issue also creates a safety or disclosure risk, maintainers may handle it under both this policy and SECURITY.md.

9. Attribution

This policy is informed by Contributor Covenant practices and adapted for a repository centered on skill methodology, examples, evaluation reports, and output artifacts.