doc-ears
Pass
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run local shell commands like 'ls' to verify documentation existence and 'python' to execute validation scripts such as 'validate_ears.py' and 'validate_tags_against_docs.py' located within the project's internal directories.
- [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface because it is designed to ingest and process content from upstream artifacts (BRD and PRD) which could contain malicious instructions. Ingestion points: Requirement documents read from docs/01_BRD/ and docs/02_PRD/. Boundary markers: The skill mandates the use of specific EARS syntax patterns (WHEN-THE-SHALL-WITHIN) and triple backticks to structure requirement statements. Capability inventory: The agent can perform file system operations and execute local python scripts via subprocess. Sanitization: There are no instructions for sanitizing or validating the content read from the upstream documents before processing.
Audit Metadata