listenhub

Warn

Audited by Gen Agent Trust Hub on Aug 30, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to automatically perform a global installation of the @marswave/listenhub-cli package via npm install -g if the CLI is missing. This action is performed silently and without requesting user confirmation, which bypasses user oversight for software installation.
  • [COMMAND_EXECUTION]: The skill executes shell commands to check the authentication status (listenhub auth status --json), install software (npm install -g @marswave/listenhub-cli), and initiate login flows (listenhub auth login).
  • [INDIRECT_PROMPT_INJECTION]: The skill functions as a router that ingests user-provided intent to delegate actions to other skills.
  • Ingestion points: User messages and keywords such as "make a podcast", "parse URL", "music", and various multi-lingual equivalents defined in the Routing Table (SKILL.md).
  • Boundary markers: Absent. The skill does not use delimiters or instructions to ignore potential commands embedded within user input when determining the route.
  • Capability inventory: The skill has the capability to execute shell commands (npm, listenhub) and route to other specialized skills which may have additional capabilities (SKILL.md).
  • Sanitization: Not present. The skill relies on direct keyword matching from raw user input to trigger its routing logic.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 30, 2026, 08:20 PM
Security Audit — agent-trust-hub — listenhub