orchestrator
Pass
Audited by Gen Agent Trust Hub on Aug 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill defines clear authority boundaries, explicitly stating that environment variables, tools, or repository instructions do not grant delegation rights without specific task authorization in
SKILL.md. - [SAFE]: It implements defensive role isolation by prohibiting terminal agents from accessing orchestration tools or lifecycle controls in
references/conventions.md, effectively preventing privilege escalation within a multi-agent environment. - [INDIRECT_PROMPT_INJECTION]: The skill manages a potential injection surface where sub-agents could provide deceptive reports in an attempt to influence the orchestrator.
- Ingestion points: The Orchestrator role reads and evaluates reports produced by terminal workers as described in
SKILL.mdand the brief template. - Boundary markers: Task boundaries are enforced via structured briefs in
references/brief-template.mdwhich limit worker scope and explicitly forbid spawning or delegation. - Capability inventory: The Orchestrator has the authority to launch new agents and verify artifacts, which are restricted capabilities.
- Sanitization: The instructions require the Orchestrator to independently verify all claims against the source or artifact rather than trusting the sub-agent's report text, providing a manual sanitization step.
Audit Metadata