agent-team-orchestration
Pass
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill contains only informational markdown files and metadata. There are no scripts, binaries, or automated installation processes.
- [PROMPT_INJECTION]: The documented workflow creates an indirect prompt injection surface by design, as agents are instructed to read and act upon data produced by other agents in shared directories.
- Ingestion points: As described in references/communication.md, agents ingest data from shared directories such as /shared/specs/ and /shared/artifacts/.
- Boundary markers: No explicit delimiters or instructions to ignore embedded commands are defined in the Spawn Prompt Template.
- Capability inventory: The framework references the use of sessions_spawn and sessions_send for inter-agent coordination.
- Sanitization: The documentation does not prescribe methods for sanitizing or validating agent-generated content before it is consumed by other agents.
Audit Metadata