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 ingests external feature specifications and requirements to generate formatted implementation plans stored in the project's documentation. There is a risk that malicious instructions provided in the requirements could be persisted in these plan files and later executed by downstream subagents. Ingestion points: User-provided specifications or task descriptions used to populate the plan template. Boundary markers: The skill does not implement delimiters or safety instructions to distinguish between untrusted input and the plan's structural logic. Capability inventory: The skill performs file system writes to the docs/plans/ directory. Sanitization: No validation or sanitization is performed on the input content before it is written into the plan document.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 02:28 AM