doc-prd-reviewer
Warn
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the agent to execute bash commands such as sha256sum and sed for document integrity verification as specified in Section 10.3. These commands are constructed using file paths and identifiers extracted directly from the processed documents.
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection via untrusted PRD and BRD files. 1. Ingestion points: Reads markdown files from local project directories. 2. Boundary markers: No explicit delimiters or safety instructions are provided for ingested content. 3. Capability inventory: Shell execution and filesystem write operations. 4. Sanitization: The instructions lack any requirement to sanitize or validate extracted strings before shell interpolation.
Audit Metadata