team-agent-orchestration
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines an orchestration workflow where a primary agent reads and acts upon data generated by other agents or external sources, such as handoff artifacts, review notes, and logs. This introduces a surface area for indirect prompt injection if the content of these processed files contains instructions designed to manipulate the orchestrating agent.
- Ingestion points: The skill explicitly references reading 'handoff.md' files, 'review notes', and 'eval reports' in the Operating Model and Agent Kanban sections.
- Boundary markers: While the skill utilizes structured JSON cards and 'merge gates' to track progress, it does not specify the use of prompt delimiters or explicit instructions to the LLM to ignore natural language commands embedded within the data it processes.
- Capability inventory: The skill body focuses on workflow management and does not contain shell commands, network operations, or filesystem write instructions itself, though it orchestrates agents that may possess these capabilities.
- Sanitization: The instructions do not include steps for sanitizing, escaping, or validating the content of external handoff files before they are ingested by the agent.
Audit Metadata