Instant messaging is still the primary interface for real conversations; this project treats IM as the canonical surface for agentic LLMs and operationalizes the whole stack. The core insight: instead of reimplementing adapters and ops for each platform, assemble a single pipeline that binds LLM providers, tool-calling, and plugin skills to many IM frontends while keeping production concerns (auth, rate limits, monitoring) central.
What Sets It Apart
- Unified multi-platform adapters: ship the same bot logic to Discord, Telegram, Slack, WeChat/WeCom, QQ, LINE, Lark, DingTalk, KOOK and Matrix bridges without rewriting platform code — one codebase for many frontends.
- Agent- and developer-friendly features: built-in MCP (Model Context Protocol) endpoint for agent-driven management, an in-repo skills folder for programmatic agents, and first-class support for tool calling and streaming outputs.
- Rich integrations and extensibility: connectors for major LLMs (OpenAI, Anthropic, Google Gemini, DeepSeek, Ollama, LM Studio, etc.), LLMOps (Dify), and many gateway/GPU platforms; extensible plugin marketplace and event-driven plugin architecture.
- Production-grade operations: access control, rate limiting, sensitive-word filtering, comprehensive monitoring, exception handling, and ready-to-run deployment options (one-line launch, Docker Compose, cloud templates).
Who It's For
Great fit if you need a single, extensible platform to run LLM-driven agents across multiple IM platforms and want built-in ops features for production deployment. Ideal for engineering teams that must integrate LLMs with existing workflows, RAG/knowledge bases, or automation tools (n8n/Dify/Langflow).
Look elsewhere if you only need a single-platform chatbot with minimal ops (a lightweight SDK) or if you prefer a hosted SaaS that hides infra and custom integrations; running and customizing the platform requires developer effort and some operational know-how.
Where It Fits
Positioned between low-level SDKs (which force per-platform work) and closed-hosted bot SaaS: it offers deep customization and self-hosting with enterprise-oriented controls and an extensible plugin/skill ecosystem. Use it when cross-platform reach, agent capabilities, and production readiness matter more than zero‑maintenance hosting.
