project-narrative
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from repository sources like commit messages and issue descriptions to construct a project narrative. This represents a surface for indirect prompt injection where instructions hidden in the source data could attempt to influence the agent's output.\n
- Ingestion points:
SKILL.md(Procedure step 1: 'Read... recent commits and open issues').\n - Boundary markers: Absent; there are no explicit delimiters or instructions to ignore embedded prompts in the source data.\n
- Capability inventory: File system write (
NARRATIVE.md).\n - Sanitization: Absent for injection; however, the skill includes 'Verify' steps to ensure no secrets or personal data are included.\n- [COMMAND_EXECUTION]: The skill instructs the agent to use the GitHub CLI to check repository visibility settings.\n
- Command:
gh repo view --json visibility,isPrivate.\n - Context: Used to determine whether the narrative should follow public or private disclosure rules, enhancing the skill's security posture.
Audit Metadata