orchestrate-omp
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and integration of outputs from multiple subagents (e.g., via agent:// and history:// URIs). This creates a vulnerability surface where the root coordinator could process malicious instructions embedded in subagent results if a child agent is compromised.\n
- Ingestion points: Subagent results are read from agent:// and session history via history://.\n
- Boundary markers: The skill requires standardized status markers (PASS, ISSUES, BLOCKED) for handoffs, providing some structure to outputs.\n
- Capability inventory: The coordinator has the ability to spawn new subagents using the task tool and access local files via local:// URIs.\n
- Sanitization: The skill mandates that the coordinator must 'Inspect each claimed artifact,' 'reconcile conflicts,' and 'Run the affected validation and real user path from the root' before acceptance, which acts as a verification layer to prevent unreviewed code or data from being integrated.
Audit Metadata