release
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from git commit messages and source code diffs to generate documentation and changelogs. A malicious contributor could embed instructions within commit messages to override the AI's changelog drafting logic or documentation updates.
- Ingestion points: Shell outputs from
git logandgit diff(referenced in Phase 1 and Phase 2). - Boundary markers: Absent; the AI is not instructed to treat the git history as untrusted or to use specific delimiters to isolate commit content.
- Capability inventory: Writing to
CHANGELOG.md, editing source files to remove debug artifacts, and invoking external project-specific skills. - Sanitization: Absent; the skill lacks validation or sanitization of commit messages before they are processed into draft release notes.
Audit Metadata