AI Detect MCP
Run a local text detector from a compatible agent and inspect sentence-level scores or model readiness.
Overview
AI Detect exposes a local AI-text detector through a command-line interface and an MCP server. The project documents scoring pasted text or a local file and returning sentence-level diagnostics. Treat those scores as signals to inspect when reviewing a draft. The first detection can wait for model weights to finish downloading. The status tool reports whether the model is ready, downloading or absent, which helps distinguish setup time from a failed connection.
Getting started
Install the project using its README. Register ai-detect-mcp with your client, or use the documented Python module invocation if the command is unavailable on your path.
Frequently asked questions
How do I tell whether the model is ready?
The documented get_model_status tool reports whether the weights are ready, still downloading or absent.
Can the detector read a local draft?
Yes. The MCP tools include a local file scoring operation, so the draft does not need to be pasted into the conversation.