Skip to content

Releases: GenAI-Security-Project/GenAI-Security-Advisor

Release list

v1.0.1 - Separate author/organization metadata

v1.0.1 - Separate author/organization metadata Latest
Latest

Choose a tag to compare

View all tags
@sclintonowasp sclintonowasp released this 10 Aug 15:36
v1.0.1
953c0bf

Changed

  • Split the combined "Scott Clinton (OWASP GenAI Security Project)" author string into separate, distinct metadata fields everywhere it appeared:
    • SKILL.md frontmatter: author: Scott Clinton + metadata.organization: OWASP GenAI Security Project
    • .claude-plugin/marketplace.json: plugin author is now person-only (organization was already separately represented via the top-level owner field)
    • gemini-extension.json: added standalone author/organization fields (not part of Gemini's documented extension schema, so likely inert rather than something the installer reads -- included for consistency/transparency)

Patch release, no behavioral change. Full history in CHANGELOG.md.

Assets 2
Loading

v1.0.0 - Initial public release

v1.0.0 - Initial public release

Choose a tag to compare

View all tags
@sclintonowasp sclintonowasp released this 10 Aug 15:29
v1.0.0
e790eec

Initial public release. Initiative lead: Scott Clinton, OWASP GenAI Security Project.

Added

  • SKILL.md: skill instructions grounding GenAI/LLM/agentic security guidance in OWASP GenAI Security Project research (LLM Top 10, Agentic Top 10, Data Security, MCP security, red-teaming, incident response, governance).
  • corpus/: 19 cataloged resources across 6 categories (15 vendored, 4 linked pending license confirmation) -- see corpus/MANIFEST.yaml.
    • The published OWASP Top 10 for LLM Applications 2026.
    • The finished OWASP Top 10 for Agentic Applications 2026 (ASI01-10).
    • The GenAI Security Crosswalk: 41 risks mapped to 25 compliance frameworks.
    • Data security, MCP security, red-teaming, and governance/COMPASS resources.
  • Cross-platform discovery for Claude Code, OpenAI Codex CLI, GitHub Copilot, and Gemini CLI, plus install manifests (.claude-plugin/marketplace.json, gemini-extension.json).
  • scripts/validate_corpus.py + CI validation on every PR touching corpus/.

Licensing

  • Repo-native content (SKILL.md, scripts/, docs): Apache-2.0.
  • Vendored corpus/ content: retains its original license per corpus/MANIFEST.yaml entry -- predominantly CC BY-SA 4.0, confirmed against each source document before vendoring.
  • Four resources are status: linked (not vendored) specifically because their license could not be confirmed: the Governance Checklist, LLM Exploit Generation report, and GenAI Incident Response Guide.

Full details in CHANGELOG.md.

Loading