Skip to content

Security: taiyuli3131/intent-quality

Security

SECURITY.md

Security Policy

intent-quality is designed as a local-first tool. It should not upload local project data, profile data, contribution packages, public-sample adoption decisions, or diagnosis reports automatically.

Reporting Security Issues

If this repository is published on a hosting platform, please report security issues through the repository's private security advisory channel when available. If no private channel exists yet, avoid posting sensitive details publicly; open a minimal issue asking for a private contact path.

Security Boundaries

The project should preserve these boundaries:

  • Public samples are untrusted by default.
  • Public sync may fetch or stage candidates, but adoption remains user-confirmation gated.
  • Contribution package generation is not submission authorization.
  • Profile memory suggestions are pending, project-local, and review-only.
  • check, eval, suggest list, and contribution review flows should remain read-only.
  • No command should silently upload, publish, or apply local state changes.

Sensitive Data

Do not include the following in issues, examples, fixtures, public candidates, or contribution packages:

  • API keys, tokens, credentials, or private keys.
  • Private filesystem paths.
  • Personal email addresses or identifiers.
  • Proprietary conversation excerpts.
  • Customer, employer, or repository data that has not been anonymized.

Verification

Before publishing or reviewing a security-sensitive change, run:

python -m intent_quality.cli check
python -m compileall intent_quality

There aren't any published security advisories