orca-cli
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local CLI tool to perform operations on worktrees and terminals. It includes logic to correctly resolve the executable path based on environment variables and system type, specifically warning against triggering the GNOME screen reader on Linux systems.
- [DYNAMIC_EXECUTION]: The skill uses a command (
ORCA skills get orca-cli) to fetch operational instructions at runtime. This design pattern ensures that the agent uses instructions compatible with the specific version of the Orca tool installed on the host system. - [INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it treats the output of a local command as instructions. However, the source is a local binary intended for this purpose, and the ingestion is a standard part of the tool's integration.
- Ingestion points: Command output from the resolved Orca executable.
- Boundary markers: None specified in this discovery stub.
- Capability inventory: Terminal management, worktree operations, and artifact management through the CLI.
- Sanitization: None; the output is intended to be loaded directly as a guide for the agent.
Audit Metadata