voicemode-connect

Fail

Audited by Gen Agent Trust Hub on Feb 14, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (HIGH): The installation instructions require running npx -y mcp-remote https://voicemode.dev/mcp. This pattern downloads and executes code from the npm registry and connects to a remote definition on a non-trusted domain.
  • [REMOTE_CODE_EXECUTION] (HIGH): By using npx to pull a remote MCP configuration from https://voicemode.dev/mcp, the skill facilitates the execution of arbitrary remote code on the user's local environment. The domain voicemode.dev is not within the defined trusted sources.
  • [DATA_EXFILTRATION] (MEDIUM): The skill's primary function is to route voice conversations through a third-party cloud platform (voicemode.dev). This constitutes a risk of data exposure for any sensitive information shared during voice sessions.
  • [INDIRECT_PROMPT_INJECTION] (MEDIUM): High-risk attack surface detected in the converse tool.
  • Ingestion points: Untrusted voice data is received from external clients (iOS/web) and processed by the agent.
  • Boundary markers: No explicit delimiters or instructions to ignore embedded commands in the voice transcripts are mentioned.
  • Capability inventory: The skill enables network communication and local command execution via the mcp-remote wrapper.
  • Sanitization: There is no evidence of sanitization for the converted text before it is presented to the agent's context.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 14, 2026, 05:16 AM