skills/tw93/kaku/product-docs/Gen Agent Trust Hub

product-docs

Pass

Audited by Gen Agent Trust Hub on Jun 28, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: Executes local Python scripts (scripts/highlight.py) and standard library modules (http.server) for the purpose of syntax highlighting and verifying the visual rendering of documentation pages in a local environment.
  • [SAFE]: Accesses local source code files (e.g., config/src/config.rs) and shell scripts to ensure documentation accuracy, which is consistent with its stated purpose of preventing documentation drift.
  • [SAFE]: Implements a security boundary by explicitly prohibiting automatic git commits or pushes to the production branch (vercel), requiring a manual diff review and explicit maintainer approval.
  • [SAFE]: Potential Indirect Prompt Injection surface identified:
  • Ingestion points: Reads content from git log, gh release list, AGENTS.md, CLAUDE.md, and various website source files (.html).
  • Boundary markers: Lacks explicit structural delimiters for external content, but uses instructional directives to "verify, never infer" against the codebase.
  • Capability inventory: Includes file writing (HTML/Markdown), local command execution (Python, Git, GH CLI), and local network serving for preview purposes.
  • Sanitization: Requires a manual human-review checkpoint ("Hand off") where the agent must present a diff for inspection before finalization.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 28, 2026, 05:49 AM
Security Audit — agent-trust-hub — product-docs