# opensquilla/opensquilla

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

Repository: https://github.com/opensquilla/opensquilla
Canonical: https://ross.abutalabs.com/products/opensquilla
Homepage: https://opensquilla.ai
Language: Python
License: Apache-2.0
License Family: permissive
Topics: agent, ai, ai-agents, deep-learning, foundation-models, llm, mcp, memory, openclaw, python, skills
Last push: 2026-08-26T20:25:00+00:00

## Health v2 (maintenance only)
Score: 81/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 99, longevity 8
- inputs: {"age_days": 119, "days_push": 7, "days_rel": 8, "gap_med": 7.5, "n_releases_24m": 11}
- flags: young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 6687, forks 528 (observed 2026-08-28T04:09:47.665672+00:00)

## What it is
OpenSquilla is a token-efficient, microkernel AI agent that routes each conversation turn to the cheapest capable LLM via a local model router, with persistent memory, a layered sandbox, built-in web search, and on-device embeddings. It exposes a shared turn loop across CLI, Web UI, Desktop, and chat channels, and supports 20+ LLM providers through a pluggable provider layer.

## Use cases
- run an AI agent in my terminal with lower token costs
- route tasks to cheaper LLM models automatically
- build an agent with persistent memory and web search
- connect one agent to OpenAI, Anthropic, Ollama and other providers
- use an AI assistant across CLI, web UI and chat channels
- cut LLM API costs without losing answer quality
- self-host an AI agent with a secure sandbox

## When to choose
- you want to minimize LLM token spend while keeping quality
- you need one agent accessible from CLI, web, desktop, and chat channels
- you want multi-provider LLM support without changing config
- you need persistent memory, scheduling, and skills in a local agent

## When to avoid
- you need a lightweight library to embed in your own app rather than a standalone agent
- you require a single fixed model with no routing overhead
- you need non-Python environments (Python 3.12+ required)

## Facets
- artifact type: application
- maturity: active
- function: agent-framework, llm-inference, rag, mcp, chatbot, search-engine, cli, gui
- domain: artificial-intelligence, large-language-models, developer-tools
- platform: python, cross-platform, cli
- tags: token-efficiency, model-routing, microkernel-agent, persistent-memory, multi-provider, sandbox, local-embeddings, cost-optimization, ai-agents, automation, desktop, web-server

## Member repositories
- opensquilla/opensquilla (main) score 81

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:47.665672+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:42:59.600670+00:00, confidence not recorded.
  - readme: https://github.com/opensquilla/opensquilla (fetched 2026-08-28T04:09:47.665672+00:00, sha 93f187d46c49)
  - homepage: https://opensquilla.ai (fetched 2026-08-29T08:39:10.866170+00:00, sha b48f887c9091)
- Data as of 2026-08-30T08:39:29.467469+00:00.