product-roadmap

Pass

Audited by Gen Agent Trust Hub on Sep 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external GitHub repositories, including issue titles, pull request bodies, and documentation, which could contain malicious instructions.
  • Ingestion points: Fetches external data using gh issue list, gh pr list, and local file reads of README and documentation files.
  • Boundary markers: The skill uses a conceptual reasoning framework to distinguish external testimony ('shabda') from verified reality, but it lacks technical delimiters to prevent the agent from executing instructions found within the ingested text.
  • Capability inventory: The skill is capable of executing shell commands via the GitHub CLI, writing multiple files to the local disk (roadmap.md, roadmap.html, etc.), and spawning subagents.
  • Sanitization: While the output HTML template includes a basic escaping function (esc) for rendering, the intermediate reasoning steps do not explicitly define sanitization for the agent's internal context.
  • [COMMAND_EXECUTION]: The skill utilizes the gh (GitHub) CLI to perform repository operations, including cloning repositories (gh repo clone) and querying the GitHub API for project metadata.
  • [CREDENTIALS_UNSAFE]: The skill instructions direct the agent to analyze the "configuration surface" of the target repository, specifically including environment variables (env). This involves scanning for sensitive configuration keys to map the application's architecture.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 6, 2026, 01:35 PM
Security Audit — agent-trust-hub — product-roadmap