om-pipeline-retro
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external tracker sources (PR descriptions, reviews, and comments) and a repository-local extension file.
- Ingestion points: Data is gathered via tracker operations and an extension file (".ai/skills/om-pipeline-retro/SKILL.md") as described in "references/agentic-setup.md".
- Boundary markers: Instructions in "references/agentic-setup.md" explicitly mandate treating external content as data and ignoring embedded directives. The classification script "references/classify-runs.sh" filters out common quoting and fencing patterns.
- Capability inventory: The skill performs local script execution, tracker API calls, and companion skill invocation.
- Sanitization: Input is validated via regex for dates and processed using "jq" for structured data handling.
- [COMMAND_EXECUTION]: The skill executes a local shell script to perform deterministic classification of pull request data.
- Evidence: "SKILL.md" step 4 uses "sh references/classify-runs.sh" to process the JSON evidence gathered from the tracker.
Audit Metadata