中文 · Documentation · Apache-2.0 License

AOS

AOS means Agent Operating System: a Web-first, multi-tenant workspace for building useful Agent workflows around one recoverable session. It brings conversation, evidence-based research, data questions, memory, files, Skills, MCP, task recovery, and external Bot channels into one product surface.

Project status: AOS is an active source release under regression testing. This repository contains the source tree, documentation, tests, and packaging scripts. Prebuilt AOS Offline archives are intentionally kept out of Git history and will be attached to GitHub Releases only when a build is ready. Some integrations require provider credentials and external services.

What AOS is for

• One recoverable Agent session for normal questions, tool use, long-running work, and follow-up context.
• Deep research that gathers and checks evidence with bounded execution and a usable fallback answer.
• Data work with datasource-aware NL2SQL, semantic retrieval, SQL knowledge, and attribution workflows.
• Mobile Agent access through Bot Gateway adapters for supported chat platforms and notification delivery.
• Progressive engineering design that turns a product request into a reviewable core design, then an editable implementation plan and task breakdown for an external coding Agent.
• Skills and MCP as optional extensions that can be installed, inspected, and governed.

The project is designed to make Agent work inspectable and recoverable. It does not claim that every provider adapter or every workflow is production-ready on every platform yet.

The menu map is a product map, not a promise that every capability is enabled in every deployment. Tenant permissions, provider support, network access, and feature maturity are shown explicitly in the linked documentation.

Architecture at a glance

Web UI / external Bot
        |
        v
   Rust WebServer  --->  SQLite platform state
        |
        +--> Agent runtime and recoverable sessions
        +--> Model providers and optional web-search extensions
        +--> Datasources, Skills, MCP, and repository adapters

The platform database is SQLite and belongs to the AOS control plane. Business databases are configured as external NL2SQL datasources; they are not required to boot AOS. A single AOS process owns one local data directory.

Try the source tree

The shortest supported path for contributors is Docker:

./scripts/generate-env.sh
docker compose up --build

Open http://localhost:3000 and complete setup. Add an enabled chat-scoped API key in System -> API Keys before asking the model to do provider-backed work.

For native development:

./scripts/setup-environment.sh --check
./scripts/aos-start.sh

Read the deployment guide before exposing AOS beyond localhost:

• Installation
• Open-source deployment
• Complete test guide

Repository map

• rust/ - Rust workspace, API, runtime, Agent orchestration, and WebServer.
• webui/ - React and Vite application.
• docs/ - architecture, deployment, security, integration, and test documentation.
• eval/ and examples/ - deterministic fixtures and small integration examples.
• scripts/ - setup, start/stop, upgrade, and packaging helpers.

Development checks

cd rust
cargo fmt --all
cargo check -p web-server
cargo test --workspace --all-features

cd ../webui
npm ci
npm run typecheck
npm run build:ci

Targeted test guides cover the Bot Gateway, data attribution, NL2SQL configuration, and the progressive engineering design workflow.

Configuration boundaries

• Model credentials are stored per tenant through the API Keys surface and are never committed to the repository.
• Search providers and MCP servers are optional extensions. The built-in fallback path remains the default when an enhancement is not configured.
• Skills are treated as untrusted input during installation and are subject to safety checks and permission boundaries.
• Long-running tasks persist state and events so a client reconnect can recover context; external IM delivery still depends on each platform's API policy.

Documentation

• Architecture
• Model capability profiles
• Long-running tasks and mobile operations design
• Bot capability contract
• Bot Gateway enterprise setup
• NL2SQL design
• Engineering design center
• Readability specification
• Security policy
• Contributing

Release artifacts

AOS Offline is a separately built distribution for macOS, Linux, and Windows x64. Release archives are not committed to this repository. Maintainers build and validate an archive, then publish it under GitHub Releases with its checksum and upgrade notes. Until that happens, use the source workflow above.

License

AOS is licensed under the Apache License 2.0. Third-party and upstream attribution notices are in NOTICE.md.