orch-pipeline
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill orchestrates a development workflow that ingests data from external sources, which creates a surface for indirect prompt injection.
- Ingestion points: Ingests external data from GitHub search results, package registries, and the Exa search engine during Phase 1, and user-provided design documents during Phase 0.
- Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded prompts for the processed external content.
- Capability inventory: The pipeline can perform repository searches via
gh, modify the local file system through TDD workflows, and perform automated code commits. - Sanitization: No automated sanitization of external content is mentioned; however, the workflow requires two manual human gates (after planning and before committing) and a specific security review phase for changes affecting sensitive areas like authentication, cryptography, or secrets.
Audit Metadata