oma-coordination

Warn

Audited by Gen Agent Trust Hub on May 3, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands via the 'oma' CLI tool and local scripts to coordinate tasks between different agents.
  • Evidence: 'SKILL.md' contains the command 'oma agent:spawn backend "task description" session-id -w ./backend &'.
  • Evidence: 'resources/examples.md' contains references to executing '.agents/skills/orchestrator/scripts/spawn-agent.sh'.
  • [PROMPT_INJECTION]: The skill facilitates indirect prompt injection by ingesting untrusted user data (task descriptions) and passing it directly to sub-agents or CLI tools without sanitization.
  • Ingestion points: User-provided 'task description' in 'SKILL.md' and task descriptions in 'resources/examples.md'.
  • Boundary markers: Absent. User input is interpolated directly into shell command arguments.
  • Capability inventory: Shell execution capabilities via 'oma' and 'spawn-agent.sh' across all scripts.
  • Sanitization: Absent. The instructions do not provide any guidance on validating or escaping user-provided task descriptions before they are used in CLI commands.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 3, 2026, 01:32 AM