skills/huangrx6/specforge/sf-steering/Gen Agent Trust Hub

sf-steering

Pass

Audited by Gen Agent Trust Hub on Jun 25, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute several local Node.js scripts (e.g., codebase-index.mjs, wiki-hydrate.mjs, wiki-quality.mjs) located in the project's .specforge/core/scripts/ directory to perform analysis and documentation updates.
  • [EXTERNAL_DOWNLOADS]: The skill includes logic to detect missing code intelligence providers and facilitates the installation of external tools like CodeGraph through user prompts or agent-assisted execution.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8) because it ingests and summarizes content from arbitrary codebases to generate persistent documentation. (1) Ingestion points: Reads source code, manifests, and documentation from the project directory. (2) Boundary markers: No specific delimiters or safety warnings are provided for processing untrusted code. (3) Capability inventory: Executes local scripts and writes to the filesystem. (4) Sanitization: No explicit validation or filtering of codebase content is mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 25, 2026, 08:20 AM
Security Audit — agent-trust-hub — sf-steering