analyze

Pass

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it ingests and processes untrusted data from the user's codebase and raw ideas to generate documentation.
  • Ingestion points: The skill scans the workspace structure, reads configuration files (e.g., .cursorrules, CLAUDE.md), and processes user-provided feature requests (SKILL.md, Step 2 & 3).
  • Boundary markers: The skill implements mandatory human review gates between every phase (Step 2-7) to verify generated artifacts before proceeding, which serves as a mitigation against automated exploitation.
  • Capability inventory: The skill performs file system write operations to save markdown artifacts (context-map.md, prd.md, etc.) based on the processed data (SKILL.md, Output section).
  • Sanitization: No explicit sanitization or instruction-filtering of the ingested codebase content is mentioned in the pipeline process.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 14, 2026, 04:41 AM