plan-review-gate

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) due to its core function of reading and processing externally provided or generated plan files.
  • Ingestion points: The skill reads 'plan_file_path' and 'user's original request' content directly into the agent's context for evaluation.
  • Boundary markers: There are no explicit instructions or delimiters (e.g., XML tags or block quotes with 'ignore instructions' warnings) to prevent the LLM from following instructions embedded within the plan being reviewed.
  • Capability inventory: The skill allows access to powerful tools including Bash, Read, Grep, and Glob, which could be misused if the LLM obeys instructions contained within a malicious plan.
  • Sanitization: The workflow lacks any sanitization or validation of the plan content prior to it being processed by the LLM for the review checks.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 07:17 PM