release
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external content including Pull Request descriptions and commit messages to generate changelogs and release notes. This creates a surface where malicious instructions embedded in PR metadata could potentially influence the agent's output.\n
- Ingestion points: Pull Request source content and commit history referenced in SKILL.md.\n
- Boundary markers: The instructions do not specify explicit delimiters or warnings for the agent to ignore instructions embedded within the ingested external texts.\n
- Capability inventory: The skill has the ability to write to files (CHANGELOG.md, .release-notes.md), execute versioning tools (mise), and perform Git operations (git tag, git push).\n
- Sanitization: No evidence is provided for automated sanitization or filtering of the ingested PR or commit text before it is processed by the model.\n- [COMMAND_EXECUTION]: The skill executes several command-line tools to automate the software release workflow.\n
- Evidence: The skill uses git for branch management and tagging, and the mise tool manager for running defined release tasks and documentation refreshes.\n- [EXTERNAL_DOWNLOADS]: The skill identifies and retrieves external resources and tools required for the release process from well-known sources.\n
- Evidence: Fetches upstream skill definitions from the grafana/ai-kit GitHub repository using the GitHub CLI (gh) and installs the svu (Semantic Version Util) tool using the mise manager.
Audit Metadata