blueprint-derive-tests

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFE
Full Analysis
  • [DYNAMIC_CONTEXT_INJECTION]: The skill uses dynamic execution syntax (!command) to gather project metadata at load time. The commands used, such as git rev-parse and find, are restricted to local repository inspection and align with the skill's functional requirements.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from git commit logs. * Ingestion points: Git commit messages are retrieved using git log in SKILL.md (Steps 4 and 5). * Boundary markers: The instructions do not define delimiters for commit message content. * Capability inventory: The skill performs file writes and manifest updates using Bash, jq, and the Write tool (Step 7). * Sanitization: No specific sanitization or validation of commit message text is implemented. However, the risk is localized to the agent's interpretation of the repo history.
  • [COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform git operations, directory management, and manifest manipulation. These actions are standard for development workflow automation.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 04:17 PM
Security Audit — agent-trust-hub — blueprint-derive-tests