diff --git a/README.md b/README.md index 722de67c..3926e177 100644 --- a/README.md +++ b/README.md @@ -1,35 +1,100 @@ -# ANOLISA — An Agentic OS Implementation +# ANOLISA, An Agentic OS Implementation + +![License](https://img.shields.io/github/license/alibaba/anolisa) +![Last Commit](https://img.shields.io/github/last-commit/alibaba/anolisa) +![Repo Stars](https://img.shields.io/github/stars/alibaba/anolisa?style=social) [中文版](README_CN.md) -ANOLISA, the Agentic evolution of Anolis OS, aims to deliver the -best-practice implementation of Agentic OS — a server-side operating -system built for AI Agent workloads. +ANOLISA, the Agentic evolution of Anolis OS, is a server-side operating system stack built for AI agent workloads. It combines terminal tooling, security controls, observability, token-efficiency tooling, and reusable operating skills into a single Agentic OS implementation. > **A**gentic **N**exus **O**perating **L**ayer & **I**nterface **S**ystem **A**rchitecture +## Table of Contents +- [Why ANOLISA](#why-anolisa) +- [Components](#components) +- [Quick Start](#quick-start) +- [Build from Source](#build-from-source) +- [Repository Layout](#repository-layout) +- [Documentation](#documentation) +- [License](#license) + +## Why ANOLISA +ANOLISA is aimed at teams building or operating AI agents on servers. It packages several core capabilities that are usually spread across separate tools: + +- AI-native shell workflows for coding and task execution +- sandboxing and security enforcement for agent actions +- observability for agent processes and LLM usage +- token optimization helpers for lower-cost inference +- reusable OS-oriented skills for operations workflows + ## Components | Component | Description | |-----------|-------------| | [Copilot Shell](src/copilot-shell/) | AI-powered terminal assistant for code understanding, task automation, and system management. Built on [Qwen Code](https://github.com/QwenLM/qwen-code). | -| [Agent Sec Core](src/agent-sec-core/) | OS-level security kernel — system hardening, sandboxing, asset integrity verification, and security decision-making. | -| [AgentSight](src/agentsight/) | eBPF-based observability for AI Agents — zero-intrusion monitoring of LLM API calls, token consumption, and process behavior. | -| [Token-less](src/tokenless/) | LLM token optimization toolkit — schema/response compression and command rewriting to reduce token consumption. | +| [Agent Sec Core](src/agent-sec-core/) | OS-level security kernel for system hardening, sandboxing, asset integrity verification, and security decision-making. | +| [AgentSight](src/agentsight/) | eBPF-based observability for AI agents, including process behavior and LLM API monitoring. | +| [Token-less](src/tokenless/) | Token optimization toolkit for schema compression, response compression, and command rewriting. | | [OS Skills](src/os-skills/) | Curated skill library for system administration, monitoring, security, DevOps, and cloud integration. | -See each component's README for detailed documentation. - -## Getting Started +## Quick Start +### Install packaged components ```bash -# Install all components via RPM sudo yum install copilot-shell agent-sec-core agentsight tokenless os-skills - -# Launch Copilot Shell cosh ``` -## License +### Enable sandbox hooks after installation +Run this once inside Copilot Shell to activate the bundled sandbox policy hooks: + +```text +/hooks install +``` -Apache License 2.0 — see [LICENSE](LICENSE). +## Build from Source +For a full source build: + +```bash +git clone https://github.com/alibaba/anolisa.git +cd anolisa +./scripts/build-all.sh +``` + +Useful variants: + +```bash +./scripts/build-all.sh --no-install +./scripts/build-all.sh --ignore-deps +./scripts/build-all.sh --component cosh --component sec-core +./scripts/build-all.sh --component cosh --component skills --component sec-core --component sight +``` + +For detailed dependency and platform notes, see [docs/BUILDING.md](docs/BUILDING.md). + +## Repository Layout + +```text +anolisa/ +├── src/ +│ ├── copilot-shell/ +│ ├── os-skills/ +│ ├── agent-sec-core/ +│ └── agentsight/ +├── scripts/ +│ ├── build-all.sh +│ └── rpm-build.sh +├── tests/ +│ └── run-all-tests.sh +├── docs/ +└── Makefile +``` + +## Documentation +- [Build from source](docs/BUILDING.md) +- [Chinese README](README_CN.md) +- component-specific README files under `src/` + +## License +Apache License 2.0, see [LICENSE](LICENSE).