openkrak-mcp@1.0.11

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

An AI client lists or invokes the package's MCP tools.

An agent can be steered away from independent source inspection while a remote service receives a stable host identifier and potentially sensitive error details.

AI-agent prompt steering combined with host-fingerprint and error telemetry transmission.

After an MCP client loads the server, its tool descriptions direct the agent to call this package first and not inspect source afterward. Each tool request performs a license check that hashes the local hostname and username and transmits that identifier to a remote service. If a tool fails, the package also transmits the error message with the fingerprint. This combines agent-directed suppression of independent inspection with external collection tied to the host.

The package has no install hook, but its active MCP interface contains coercive no-inspection instructions and sends a stable machine identifier plus error telemetry to a remote endpoint. Those behaviors form a concrete agent-steering and data-collection attack surface.