windsurf-release-automation

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data (git commit messages) to automate versioning and changelog generation, which creates a surface for indirect instructions to influence agent behavior.
  • Ingestion points: The skill reads git commit history and project configuration files like package.json and .releaserc.js to perform its core functions (SKILL.md, references/directory-structure.md).
  • Boundary markers: There are no documented delimiters or "ignore instructions" directives provided to the agent for the content it reads from the commit history.
  • Capability inventory: The skill has permissions for Write, Edit, and Bash operations, allowing it to modify files, create git tags, and execute release commands (SKILL.md).
  • Sanitization: The implementation guide does not specify any validation or sanitization steps for the commit messages being processed.
  • [EXTERNAL_DOWNLOADS]: The skill documentation references the use of standard package managers (npm, yarn, pnpm) to install the semantic-release tool as part of the setup process (SKILL.md, references/implementation.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:49 AM
Security Audit — agent-trust-hub — windsurf-release-automation