open-multi-agent-orchestration

Warn

Audited by Gen Agent Trust Hub on Apr 1, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs users to install the '@jackchen_me/open-multi-agent' package from the npm registry. This dependency is not maintained by the skill author or a recognized trusted organization, which introduces a potential supply chain risk.\n- [COMMAND_EXECUTION]: The framework includes a built-in 'bash' tool that allows agents to execute arbitrary shell commands. This capability significantly increases the impact of any potential compromise or malicious instruction execution.\n- [PROMPT_INJECTION]: The orchestration framework is susceptible to indirect prompt injection due to its handling of external task descriptions and user prompts. \n
  • Ingestion points: Untrusted data enters the agent context through 'orchestrator.runAgent', 'orchestrator.runTeam', or 'orchestrator.runTasks' calls defined in SKILL.md. \n
  • Boundary markers: No explicit delimiters or instructions are provided to the agents to distinguish between system orchestration commands and data-driven task content. \n
  • Capability inventory: Agents are granted high-privilege tool access, including 'bash' for shell execution and multiple file system tools ('file_write', 'file_edit', 'file_read'). \n
  • Sanitization: The documentation does not specify any validation, escaping, or filtering of external content before it is processed by the agents or their tools.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 1, 2026, 12:55 PM