implement-spec-in-pipeline
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from issue trackers and specification files to drive the development workflow.
- Ingestion points: The skill reads specifications and tickets (SKILL.md, Step 1) which are then used by various subagents.
- Boundary markers: No explicit delimiters or instructions to ignore embedded directives within the ticket content are provided.
- Capability inventory: The workflow utilizes subagents capable of code implementation, running tests via the
tddtool, and executing shell commands. - Sanitization: There is no evidence of sanitization or structural validation for the ingested ticket data before it is utilized in the prompt context of subagents.
- [COMMAND_EXECUTION]: The skill utilizes background shell execution to run the
ocrcode review utility. - Evidence: The instructions include a specific command for background execution:
bg_run: OCR_LLM_TIMEOUT=300 ocr review --from <base-branch> --to <integration-branch> ....
Audit Metadata