orchestration
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch its main usage guide and instructions at runtime by executing a local command.
- Ingestion points: The output of the
ORCA skills get orchestrationcommand (where ORCA is the resolved binary) is directly ingested into the agent's context as the authoritative guide for subsequent actions (File:SKILL.md). - Boundary markers: No boundary markers or instructions to ignore embedded malicious directives are present when the agent processes the fetched guide.
- Capability inventory: The skill environment grants access to terminal control, shell command execution, and browser automation via the Orca toolset.
- Sanitization: There is no evidence of sanitization or validation of the fetched guide content before the agent acts upon its instructions.
- [COMMAND_EXECUTION]: The skill instructs the agent to resolve a local executable and perform shell operations.
- Evidence: The skill logic resolves the executable to
orca,orca-ide, ororca-devand executes commands such asORCA status --json,ORCA orchestration task-list --json, andORCA terminal list --jsonto manage the local environment.
Audit Metadata