hyprpilot
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the execution of a custom CLI tool named
hyprpilot. It includes instructions for the agent to guide the user through a manual build and installation process (cargo buildandinstallto~/.local/bin/), ensuring the binary is not installed silently. - [INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves capturing and reading screenshots of third-party native applications.
- Ingestion points: Screen captures (PNG files) saved to
$XDG_RUNTIME_DIR/hyprpilot/shots/and window metadata retrieved viahyprpilot statusandhyprctl. - Boundary markers: None identified; the agent is instructed to "READ the PNG to verify" UI states.
- Capability inventory: The skill can perform mouse clicks, keyboard input (
type,key), and window management operations through thehyprpilotCLI. - Sanitization: There is no mechanism to sanitize or filter visual instructions that might be displayed by a malicious application to influence the agent's logic.
Audit Metadata