doc-spec-autopilot
Pass
Audited by Gen Agent Trust Hub on Apr 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands that incorporate user-controlled variables, creating a risk of path traversal or command manipulation.
- Evidence: Uses
ls docs/09_SPEC/SPEC-{NN}_*/to verify file existence, where{NN}is extracted directly from user-supplied strings. - Evidence: Executes
python ai_dev_ssd_flow/scripts/update_traceability_matrix.pyto synchronize project metadata. - [PROMPT_INJECTION]: The skill is subject to indirect prompt injection as its core logic involves reading and interpreting untrusted requirement documents.
- Ingestion points: Reads content from
REQandCTR(requirement and contract) documents as part of Phase 1. - Boundary markers: The skill instructions do not specify the use of delimiters or 'ignore' prefixes for the data ingested from external files.
- Capability inventory: The skill has the ability to read/write files, list directory contents, and execute local Python scripts.
- Sanitization: No content validation or sanitization process is described for the text extracted from the requirement documents before it influences the agent's logic.
Audit Metadata