hs-release

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes multiple local shell scripts and project-specific CLI tools, including ./scripts/release.sh, ./scripts/generate-docs-skill.sh, and uv run. These commands perform critical release actions such as tagging versions, pushing to main, and regenerating SDKs. These are standard operations for a release management skill but require the execution environment to be trusted.
  • [INDIRECT_PROMPT_INJECTION]: The skill uses an LLM-based process to summarize commit history for the changelog (uv run --directory hindsight-dev generate-changelog <version>). This introduces a surface where malicious commit messages could influence the generated documentation.
  • Ingestion points: Commit messages retrieved via git log between the previous and current versions.
  • Boundary markers: Not explicitly defined in the instructions for the LLM summarization step.
  • Capability inventory: The agent can write to documentation files (hindsight-docs/src/pages/changelog/index.md) and perform git operations.
  • Sanitization: The skill relies on the internal implementation of the generate-changelog script; no explicit sanitization of commit metadata is described in the prompt context.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 12:52 AM