Opensquilla is a free, microkernel AI agent runtime developed by Opensquilla for the mcp platform. Built primarily with Python 3.12 or newer, it functions as both an MCP client and server. It connects to a pluggable layer of numerous Large Language Model providers and integrates with the Model Context Protocol for tools and data. This program targets developers and workflow builders seeking a self-hostable solution for enterprise workloads, aiming to optimize AI token usage and provide consistent multi-channel AI interaction.

What does Opensquilla connect to?

Opensquilla connects to a pluggable layer of LLM providers including TokenRhythm, OpenRouter, OpenAI, Anthropic, Ollama, DeepSeek, Gemini, Qwen/DashScope, Moonshot, Mistral, Groq, vLLM, and LM Studio. It integrates with the Model Context Protocol for plugging in tools and data. Entry points include a Web UI, CLI, and chat channels such as Slack, Discord, WeCom, Feishu, Telegram, DingTalk, and QQ. WeCom connections specifically use WebSockets, while LLM providers interact via HTTPS APIs.

Core features and tools

Opensquilla offers a range of bundled skills for various tasks. These include capabilities for coding, GitHub integration, cron scheduling, and document generation in formats like DOCX, PPTX, XLSX, and PDF. Other skills cover summarization, tmux, weather, deep research, and bounded subagent delegation. These skills enable the AI agent to perform complex operations and interact with different systems and data types.

Who is it for?

Opensquilla is designed for developers and workflow builders seeking a self-hostable, open-source AI agent runtime. It is particularly suited for managing long-horizon enterprise workloads and for those aiming to optimize AI token usage. The program provides a consistent agent experience across various interaction channels. Skip this if you require explicit details on specific transport protocols for its server role.

How to get started

The server authenticates using an API key passed as an environment variable. For instance, OPENROUTER_API_KEY or OPENAI_API_KEY can be configured during the initial "opensquilla onboard" process.

Use cases

• Reduces AI token usage by 60-80% compared to single-model deployments through intelligent model routing and layered memory.
• Provides a unified agent experience across Web UI, CLI, Slack, and Discord with identical tool dispatch and decision logging.
• Automates document generation and deep research, creating documents in DOCX, PPTX, XLSX, and PDF formats.

Opensquilla is a versatile, free AI agent runtime that integrates with many LLM providers and offers diverse skills for enterprise use. Its multi-channel interaction and token usage optimization make it valuable for complex workflows.