vlearn-lab-authoring

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the GitHub CLI (gh api) to retrieve content from repositories. This is a legitimate use of a system tool to support the skill's core functionality of source ingestion.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external GitHub links. It mitigates this risk with instructions to treat source material as data rather than commands. Ingestion points: GitHub raw URLs and gh api calls (references/github-source-ingestion.md). Boundary markers: Instructions in SKILL.md and references/github-explanation-style.md to ignore instructions inside source content. Capability inventory: Execution of gh api and file system reads. Sanitization: Directives to ignore override instructions and focus on learning path facts.
  • [SAFE]: The skill demonstrates safe handling of credentials by explicitly forbidding the request or storage of API tokens and relying on existing local authentication states.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 04:15 PM
Security Audit — agent-trust-hub — vlearn-lab-authoring