MCP servers

Three Model Context Protocol servers that give AI agents a companion chat, a generation queue and personality matching.

All three are Python servers built on FastMCP, so any MCP-capable agent can call them as tools. The chat and queue servers generate text with OpenRouter, and every project is MIT-0 licensed.

MCP Waifu Chat

A companion character as an MCP server, with users and conversation memory.

User management: create, check, delete and count users.
Dialog history kept in SQLite, with tools to get, set and reset it.
Chat replies through OpenRouter; the model comes from an environment variable or ~/.model-openrouter.
Unit tests and a modular layout that is easy to extend.

github.com/waifuai/mcp-waifu-chat

MCP Waifu Queue

Puts text generation behind a Redis queue so many requests can run at once.

Requests are queued with python-rq and processed by worker processes that call OpenRouter.
Job status is exposed as MCP resources, so an agent can submit work and check back later.
Configured with pydantic-settings and a .env file.

github.com/waifuai/mcp-waifu-queue

MCP Traits Matcher

Personality analysis and matching over MCP.

Creates people and traits scored on friendliness and dominance.
Adding a description to a person updates their personality from the traits it mentions.
Matches people to a job description by Euclidean distance.
Stores everything in SQLite and lists it through the persons://all and traits://all resources.

github.com/waifuai/mcp-traits-matcher

Traits

Character and agent trait system, with Manim animations.

Waifu Chat API

REST API for waifu chatbot conversations with user management.

LLM Text Queue GPU

GPU-accelerated text processing pipeline.

Hermes Waifu

Live2D widget for Hermes Agent with five built-in models and expression controls.