charted-review
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests content from user-specified design documents which is then directly processed by multiple sub-agents without validation. 1. Ingestion points: The skill reads the design doc path from $ARGUMENTS[0] in SKILL.md. 2. Boundary markers: The design doc content is passed to sub-agents in full without delimiters or instructions to ignore embedded instructions. 3. Capability inventory: The skill spawns four expert sub-agents and synthesizes their output into a consolidated review with action items. 4. Sanitization: No sanitization or validation of the input file content is performed.
Audit Metadata