write-release-notes
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from sources like pull request descriptions and commit messages, which could contain malicious instructions designed to hijack the agent's behavior.\n
- Ingestion points: The skill reads commit logs, pull request context, release fragments, and third-party notes as mentioned in SKILL.md.\n
- Boundary markers: The 'Trust boundary' section in references/agent-layer.md explicitly instructs the agent to treat third-party notes and embedded commands as source material and not as authority to act.\n
- Capability inventory: The skill has the ability to read and write files for documentation updates and may execute shell commands for migration validation if specifically authorized by the user.\n
- Sanitization: The instructions tell the agent to exclude credentials and unrelated instructions from the output, though technical sanitization of the input text is not specified.\n- [EXTERNAL_DOWNLOADS]: The skill provides links to official documentation and examples from trusted organizations and well-known services, including Anthropic, OpenAI, Google, Stripe, and established open-source projects on GitHub. These references are provided for stylistic and formatting guidance and do not involve downloading executable code from untrusted or unknown sources.
Audit Metadata