orca-cli
Warn
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to invoke local command-line tools including 'orca', 'orca-ide', and 'orca-dev' based on environment state to perform system operations.
- [PRIVILEGE_ESCALATION]: The skill instructions explicitly direct the agent to 're-run it with escalated permissions' if a 'runtime_access_denied' error occurs, encouraging the agent to bypass security sandbox limitations.
- [INDIRECT_PROMPT_INJECTION]: The skill dynamically loads instructions via 'ORCA skills get orca-cli', which serves as a version-matched guide for the agent's actions. This creates a surface where external tool output can direct agent behavior. 1. Ingestion points: Data returned by the 'ORCA skills get' command as described in SKILL.md. 2. Boundary markers: No delimiters or 'ignore' instructions are provided for the ingested content. 3. Capability inventory: The agent is authorized to execute shell commands through the resolved Orca executable. 4. Sanitization: There is no evidence of sanitization or safety validation for the instructions loaded from the CLI output.
Audit Metadata