orchestration

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the orca CLI to manage terminal sessions and execute commands in those terminals (e.g., orca terminal create --command "claude", orca terminal send).
  • [PROMPT_INJECTION]: The orchestration system enables agents to send messages and inject instructions into the context of other agents using commands like orca orchestration send and orca orchestration dispatch --inject. This creates an indirect prompt injection surface where a malicious or subverted agent could influence or control others within the same Orca runtime.
  • Ingestion points: Messages are ingested via orca orchestration check --inject and task preambles via orca orchestration dispatch --inject (SKILL.md).
  • Boundary markers: No specific delimiters or safety warnings are described to distinguish between orchestration system messages and legitimate agent instructions.
  • Capability inventory: Agents can create terminals, execute shell commands, and manage a task DAG via the orca toolset.
  • Sanitization: The skill does not describe any validation or sanitization of the content exchanged between agents.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 07:10 PM
Security Audit — agent-trust-hub — orchestration