orca-cli
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill directs the agent to execute specific binaries (orca, orca-ide, orca-dev) to perform worktree and terminal operations. It includes logic to select the correct executable based on environment variables to avoid collisions with system tools.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch its core functionality and guidelines dynamically from the output of 'ORCA skills get orca-cli' rather than containing them statically. 1. Ingestion points: Command output from the Orca CLI (SKILL.md). 2. Boundary markers: Lacks delimiters or warnings to treat the fetched guide as untrusted data. 3. Capability inventory: Terminal access, worktree manipulation, and browser control. 4. Sanitization: No evidence of sanitization for the dynamic instructions.
- [DYNAMIC_EXECUTION]: The skill implements a pattern where its behavioral logic is retrieved and executed at runtime, making the full scope of its actions dependent on the external binary's output.
Audit Metadata