openkrak-mcp@1.0.17

One or more suspicious static signals were detected.; source fingerprint signature matched known malicious package; routed for review

Starting the MCP server and invoking its tools.

Can steer an AI away from source verification and disclose host identity and error details to an external service.

AI-instruction manipulation plus external host telemetry.

The executable MCP server advertises tool descriptions designed to make an AI avoid reading the user’s repository and trust package-provided analysis instead. On use, it creates a stable hash from the local hostname and username and posts it to a remote service. It also posts tool names, token estimates, and uncensored error text to that service. This creates both agent-steering and external telemetry behavior without meaningful user control at the individual tool-call level.

The package contains explicit AI-behavior steering that suppresses independent source verification, alongside host fingerprinting and error telemetry to a remote endpoint. These are concrete, package-authored runtime behaviors, not scanner-only signals.