writing-plans

Pass

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes content from research files to generate implementation plans, creating an attack surface where malicious instructions in research documents could influence agent behavior.
  • Ingestion points: Local research files matching the pattern docs/plans/YYYY-MM-DD-<topic>-research.md are ingested into the agent context.
  • Boundary markers: The instructions do not define boundary markers or delimiters to isolate content from research documents.
  • Capability inventory: The skill is capable of searching for files, writing new plan documents to the filesystem, and performing web-based research.
  • Sanitization: The skill lacks sanitization or validation logic for the external content read from research files.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 03:11 AM