TokenRhythm/opensquilla

OpenSquilla — Token-Efficient AI Agent with same budget, higher intelligence density

OpenSquilla – Token‑Efficient AI Agent

What it is – OpenSquilla is a locally‑run, microkernel‑style AI agent that routes each interaction (a turn) to the cheapest LLM that can handle the request. It combines a lightweight on‑device model router (SquillaRouter), persistent memory, a sandboxed tool layer, web‑search, and on‑device embeddings, all wrapped in a single turn loop. The same loop powers a desktop GUI, a CLI, and various chat‑channel integrations.

Key capabilities

  • Token‑efficient routing – an on‑device router selects the most cost‑effective provider (TokenRhythm, OpenRouter, OpenAI, Anthropic, Ollama, Gemini, Qwen, etc.) for each turn.
  • Unified entry points – CLI, Electron‑based desktop app, and chat channel bots all share identical tool dispatch, retry logic, and logging.
  • Pluggable provider layer – 20+ LLM back‑ends are supported without code changes.
  • Built‑in tools – persistent memory, sandboxed execution, web search, and local embeddings.
  • Extensible channels – out‑of‑the‑box support for Feishu, Telegram, DingTalk, QQ, WeCom, Slack, Discord; optional Matrix and Matrix‑E2E extras.
  • Privacy‑aware telemetry – optional network reporting with a single hard‑off switch.

Installation options

Path Who it’s for How to install
Desktop installers (macOS ARM, Windows x64) Users who want a ready‑to‑run GUI Download the .dmg or .exe from the GitHub release page and drag/run.
Quick terminal install End‑users on any OS who prefer a single command uv tool install --python 3.12 "opensquilla[recommended] @ https://github.com/TokenRhythm/opensquilla/releases/download/v0.5.6/opensquilla-0.5.6-py3-none-any.whl"
Install from source Users who want the latest main or to inspect the code Clone with Git LFS, run scripts/install_source.sh (or the PowerShell variant).
Develop from source Contributors Use uv sync --extra recommended --extra dev and run commands via uv run.

Typical workflow

  1. Run the onboarding wizard: opensquilla onboard (sets up config and secrets).
  2. Start the gateway: opensquilla gateway run.
  3. Interact via the CLI (opensquilla chat), the desktop UI, or a connected chat channel.

Maturity – The project is at version 0.5.6 (stable) with CI testing, signed Windows binaries, and a technical report published on arXiv and aiXiv (2026). It provides production‑ready installers and a documented uninstall process.

License – Apache 2.0.

Citation – If you reference OpenSquilla, cite the technical report:

OpenSquilla: Token‑Efficient AI Agent, TokenRhythm, 2026.
arXiv:2607.11399 (Agentic Routing: The Harness‑Native Data Flywheel)

Where to learn more – Full product guide (README.product.md), API docs (docs/), and privacy policy (PRIVACY.md).

Related

  • Project
    NVIDIA/OpenShell

    A private, policy-enforced runtime that lets AI agents safely read files, call APIs, and use credentials by confining each agent to an isolated sandbox and formally verifying any policy change before it goes live.

  • Project
    shibing624/agentica

    Agentica is an open‑source, Apache‑2.0‑licensed framework that lets you run LLM‑powered agents locally via a CLI, web UI, or desktop app. It supports many models, async multi‑tool execution, loss‑less context compression, automatic skill generation, and multi‑agent collaboration, with Python and TypeScript SDKs for embedding.

  • Project
    tinyhumansai/openhuman

    OpenHuman is an open‑source Rust‑based AI‑agent harness offering a desktop app and a library. It runs hundreds of agents on a cheap VPS by compressing prompts, using a tiny tool‑search model, and keeping the core in a single compiled process. Benchmarks show it’s the fastest, cheapest, and most memory‑efficient among seven comparable harnesses. Features include built‑in memory, voice agents, real‑browser/desktop control, 119 OAuth integrations, 14 messaging channels, and support for local or BYOK models. Install via pre‑built binaries or a simple script; developers embed it with a few lines of Rust code.

  • Project
    clawdotnet/openclaw.net

    OpenClaw.NET is a .NET‑native, self‑hosted AI‑agent runtime with a desktop Avalonia UI (AgentQi Companion). It runs a gateway that talks to any OpenAI‑compatible model, offers streaming, tool execution, session memory, and many integrations (REST, WebSocket, MCP, 80+ tools, channel adapters). The gateway can be compiled with NativeAOT, making it suitable for edge or on‑prem deployments. Users can start via a ready‑made desktop bundle or from source, configure the assistant through chat commands, and extend functionality with .NET plugins or reusable MetaSkill DAG workflows.

  • Project
    open-webui/open-webui

    Open WebUI is a self-hosted AI platform that provides a user-friendly interface for interacting with AI models, supporting local models via Ollama and any OpenAI-compatible API. It offers features like plugins, role-based access control, persistent memory, RAG, and more, and can run entirely offline.