Swarm Orchestration
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches the
agentic-flowpackage (version 1.5.11 or higher) from the npm registry usingnpx. - [COMMAND_EXECUTION]: Employs
npxshell commands to perform swarm initialization (swarm-init), agent creation (agent-spawn), and task management (task-orchestrate). - [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by accepting untrusted user input that influences agent swarm behavior.
- Ingestion points: The
--taskparameter innpx task-orchestrateand thegoalfield in theswarm.autoOrchestratemethod inSKILL.md. - Boundary markers: Absent; the skill does not wrap user-supplied goals or tasks in delimiters to prevent them from being interpreted as instructions.
- Capability inventory: The skill can spawn new agents, distribute tasks across a network, and manipulate shared memory stores.
- Sanitization: No input validation or filtering is specified for the task descriptions provided to the orchestrator.
Audit Metadata