openhue

Pass

Audited by Gen Agent Trust Hub on Sep 8, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches a pre-built executable binary (openhue-linux-amd64) from the project's GitHub releases page at https://github.com/openhue/openhue-cli/releases/latest/download/.
  • [COMMAND_EXECUTION]: Instructs the environment to make the downloaded binary executable using chmod +x and invokes the openhue command-line utility for all lighting operations.
  • [INDIRECT_PROMPT_INJECTION]: The skill uses natural language inputs for light, room, and scene names which are then passed directly to the command-line tool, creating a potential surface for command injection if names are not sanitized.
  • Ingestion points: User-provided strings for lights, rooms, and scenes in SKILL.md examples.
  • Boundary markers: None identified in the command templates.
  • Capability inventory: Execution of local shell commands via the openhue binary.
  • Sanitization: None described in the skill instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 8, 2026, 07:01 PM
Security Audit — agent-trust-hub — openhue