// the find
LlmKira/Openaibot
⚡️ Build Your Own chatgpt Bot|🧀 Discord/Slack/Kook/Telegram |⛓ ToolCall|🔖 Plugin Support | 🌻 out-of-box | gpt-4o
A message-queue-driven chatbot framework that connects LLM backends to Discord, Slack, Kook, and Telegram. It uses RabbitMQ for async event routing between a sender and receiver process, with a plugin system installable via pip. The project is in maintenance mode — no new features planned, and the author points you at AstrBot if you want something actively developed.
The MQ-driven architecture is genuinely well-thought-out: sender and receiver are decoupled processes, so a slow LLM response doesn't block the message intake path. Plugin distribution via pip entry points is clean — you can ship and version plugins independently without touching the core. The hook system for intercepting EventMessage gives you real extension points without forking. Multi-backend storage (Redis, MongoDB, local filesystem) means you can scale down to a single machine without ripping out the architecture.
Maintenance mode is the elephant in the room — the README explicitly says no new features are planned and recommends a different project. The dependency stack is heavy for what you get: RabbitMQ + Redis + MongoDB for a chatbot is a lot of infrastructure to operate. The OpenAI-schema-only approach means you're always one gateway proxy away from using anything non-OpenAI, which adds operational complexity. Documentation and tests are thin — the test directory has one pydantic error file and a folder of exploratory scripts, not actual test coverage.