collab-process
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes workflows for processing untrusted external data.
- Ingestion points: The operations-manual reference (Phase 1) directs the agent to ingest data from user-provided document paths, code repositories, and wiki URLs.
- Boundary markers: There are no explicit instructions or delimiters defined to separate external data from system instructions or to command the agent to ignore instructions embedded in the source material.
- Capability inventory: The skill utilizes the agent to extract workflows and implicit knowledge from sensitive locations such as code comments, commit logs, and configuration files.
- Sanitization: The instructions do not specify any validation or sanitization of the extracted content before it is used to generate manuals and training materials.
Audit Metadata