living-product-specification

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill outlines a structured approach for 'Living Product Specification,' focusing on synchronizing documentation with code changes through clear ownership and upkeep rules.
  • [COMMAND_EXECUTION]: Included Node.js validation scripts utilize built-in modules ('node:fs/promises') to perform read-only checks on documentation files. The scripts are designed for local execution to ensure structural integrity and do not perform network requests or execute untrusted shell commands.
  • [PROMPT_INJECTION]: The skill's workflow involves the agent reading Markdown files from the repository's 'docs/' directory. This constitutes an ingestion of untrusted data (Indirect Prompt Injection surface). Analysis of the evidence chain:
  • Ingestion points: 'scripts/docs.mjs' via the 'loadDocs' function, which reads the content of all '.md' files in the target directory.
  • Boundary markers: None explicitly used during file ingestion.
  • Capability inventory: Local file system read and directory listing ('readFile', 'readdir', 'stat'); no execution capabilities ('eval', 'exec') are present in the provided scripts.
  • Sanitization: None; the scripts perform regex-based structural parsing (links, headings, frontmatter) rather than content execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 12:50 AM
Security Audit — agent-trust-hub — living-product-specification