# anolisa

**Repository Path**: alibaba/anolisa

## Basic Information

- **Project Name**: anolisa
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Apache-2.0
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 2
- **Forks**: 0
- **Created**: 2026-03-31
- **Last Updated**: 2026-09-11

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

<div align="center">

<picture>
  <source
    media="(prefers-color-scheme: dark)"
    srcset="docs/images/brand/anolisa-lockup-dark.svg"
  >
  <source
    media="(prefers-color-scheme: light)"
    srcset="docs/images/brand/anolisa-lockup-light.svg"
  >
  <img
    src="docs/images/brand/anolisa-lockup-light.svg"
    alt="ANOLISA"
    width="320"
  >
</picture>

<sub>**A**gentic **N**exus **O**perating **L**ayer & **I**nterface **S**ystem **A**rchitecture</sub>

**The operating system layer for Agent workloads.**

Let Agents drive the system straight from your terminal, and strip the tool
responses that reach the model before they cost you — while keeping the Shell,
Agent framework, and sandbox you already run.

[中文版](README_zh.md) · [Website](https://agentic-os.sh/) ·
[Quick Start](https://agentic-os.sh/docs/quickstart/) ·
[User Guide](https://agentic-os.sh/docs/user-guide/) ·
[Contributing](https://github.com/alibaba/anolisa/blob/main/CONTRIBUTING.md)

[![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://github.com/alibaba/anolisa/blob/main/LICENSE)
[![Platform](https://img.shields.io/badge/Platform-Linux%20%7C%20macOS-lightgrey.svg)](https://agentic-os.sh/docs/user-guide/installation/)

</div>

---

ANOLISA is a server-side operating layer for AI Agent workloads. It addresses
three practical constraints of Agent execution: terminal entry, Token cost, and
execution environments. Keep the Shell, Agent framework, and sandbox you
already use. ANOLISA CLI provides a single installation entry point, while each
capability can be enabled independently.

**New to ANOLISA?**
[Choose your first outcome in the Quick Start →](https://agentic-os.sh/docs/quickstart/)

## Components

<table width="100%">
  <thead>
    <tr>
      <th width="340">Agent entry</th>
      <th width="340">Context efficiency</th>
      <th width="340">Runtime &amp; security</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td><strong><a href="https://agentic-os.sh/docs/user-guide/user-entrypoint/cosh-ng/quickstart/">cosh-ng</a></strong><br><sub>Shell copilot</sub></td>
      <td><strong><a href="https://agentic-os.sh/docs/user-guide/token-saving/tokenless/quickstart/">Token-less</a></strong><br><sub>Tool-output compression</sub></td>
      <td><strong><a href="https://agentic-os.sh/docs/user-guide/runtime/ws-ckpt/">ws-ckpt</a></strong><br><sub>Checkpoint and rollback</sub></td>
    </tr>
    <tr>
      <td><strong><a href="https://agentic-os.sh/docs/user-guide/user-entrypoint/os-skills/">OS Skills</a></strong><br><sub>System and DevOps expertise</sub></td>
      <td><strong><a href="https://agentic-os.sh/docs/user-guide/agent-observability/agentsight/">AgentSight</a></strong><br><sub>Trace and Token visibility</sub></td>
      <td><strong><a href="https://agentic-os.sh/docs/user-guide/runtime/skillfs/">SkillFS</a></strong><br><sub>Focused Skill views</sub></td>
    </tr>
    <tr>
      <td><strong><a href="https://agentic-os.sh/docs/user-guide/user-entrypoint/ktuner/">ktuner</a></strong><br><sub>Kernel tuning</sub></td>
      <td><strong><a href="https://agentic-os.sh/docs/user-guide/token-saving/agent-memory/">Agent Memory</a></strong><br><sub>Cross-session memory</sub></td>
      <td><strong><a href="https://agentic-os.sh/docs/user-guide/agent-security/agent-sec-core/quickstart/">Agent Sec Core</a></strong><br><sub>Sandbox and verification</sub></td>
    </tr>
    <tr>
      <td></td>
      <td></td>
      <td><strong><a href="https://agentic-os.sh/docs/user-guide/runtime/blaze/">Blaze</a></strong><br><sub>Sandbox lifecycle</sub></td>
    </tr>
  </tbody>
</table>

## What it solves

<p align="center"><strong>01 · AGENT INTERFACE</strong></p>

<h3 align="center">Let the Agent work directly in the terminal</h3>

cosh-ng is an AI-native Linux terminal: it keeps familiar Bash/Zsh behavior,
then adds an Agent that can understand intent, use tools and Skills, and ask
for approval before risky work. Shell commands and natural language share one
terminal instead of forcing users into a separate chat application.

[Get started with cosh-ng →](https://agentic-os.sh/docs/user-guide/user-entrypoint/cosh-ng/quickstart/)

<p align="center"><strong>02 · CONTEXT EFFICIENCY</strong></p>

<h3 align="center">See where Tokens go and cut waste before it reaches the model</h3>

[Token-less](https://agentic-os.sh/docs/user-guide/token-saving/tokenless/quickstart/)
removes redundancy from tool schemas and responses before they reach the model.
[Agent Memory](https://agentic-os.sh/docs/user-guide/token-saving/agent-memory/)
reuses useful context across sessions.
[SkillFS](https://agentic-os.sh/docs/user-guide/runtime/skillfs/) keeps the
current Skill view focused and makes other Skills discoverable when needed.
[AgentSight](https://agentic-os.sh/docs/user-guide/agent-observability/agentsight/)
shows where Tokens are spent.

### See an Agent run from the kernel up

On Linux, AgentSight uses eBPF to observe an Agent without changing its code.
Follow user input through model and tool calls, with Token use and sub-agent
branches in the same view.

[Open the AgentSight guide →](https://agentic-os.sh/docs/user-guide/agent-observability/agentsight/)

<table align="center" cellpadding="0" cellspacing="0">
  <tr>
    <td>
      <video
        controls
        muted
        preload="metadata"
        src="https://github.com/user-attachments/assets/ed6952c6-19da-44c7-ae35-85d753bdccf9"
      ></video>
    </td>
  </tr>
</table>

### Try Token-less with Claude Code in 3 minutes

Install Token-less and connect it to Claude Code:

```bash
curl -fsSL https://get.agentic-os.sh | bash
export PATH="$HOME/.local/bin:$PATH"
anolisa install tokenless
anolisa adapter enable tokenless claude-code
```

Restart Claude Code, run one tool-heavy task, then inspect the result:

```bash
tokenless stats summary
tokenless stats list --limit 5
```

[Open the full Token-less Quick Start →](https://agentic-os.sh/docs/user-guide/token-saving/tokenless/quickstart/)
· [Read the user manual](https://agentic-os.sh/docs/user-guide/token-saving/tokenless/user-manual/)

<table align="center" cellpadding="0" cellspacing="0">
  <tr>
    <td>
      <video
        controls
        muted
        src="https://github.com/user-attachments/assets/b372ae72-44fa-492f-9feb-e6cd137b631a"
      ></video>
    </td>
  </tr>
</table>

<p align="center">
  <sub>
    In one observed coding task, Token-less saved 317K Tokens (40.5%), based
    on AgentSight measurements.
    Results vary by workload.
  </sub>
</p>

`debug` and `trace` are dropped by the field blacklist, `metadata` as null, and
`tags` / `extra` as empty values. Compression runs between the Agent and the
model, so no Agent framework code changes. Dropped array items stay retrievable
through a `<<tokenless:KEY>>` marker, which keeps the compression reversible.

| Tool responses | Tool schemas | Full pipeline |
|----------------|--------------|---------------|
| **65.8% fewer Tokens** | **47.3% fewer Tokens** | **62.9% fewer Tokens** |
| ResponseCompressor · 46.85 µs | SchemaCompressor · 11.44 µs | 198.91 µs |

Savings apply to the tool responses entering the context, not to the whole
session bill. The [Token-less user manual](https://agentic-os.sh/docs/user-guide/token-saving/tokenless/user-manual/)
explains how to estimate the effect for a given workload.

<p align="center"><strong>03 · EXECUTION RUNTIME</strong></p>

<h3 align="center">Give every Agent execution a boundary and a way back</h3>

ANOLISA is building out the Agent execution environment:
[Agent Sec Core](https://agentic-os.sh/docs/user-guide/agent-security/agent-sec-core/quickstart/)
isolates risky operations, and
[ws-ckpt](https://agentic-os.sh/docs/user-guide/runtime/ws-ckpt/) keeps recovery
points for workspace changes.

### Catch a changed Skill before it runs

When a signed Skill changes, the Agent reports `drifted` before using it again.
A rescan records blocking findings as `deny`.

[Try the Agent demo →](https://agentic-os.sh/docs/user-guide/agent-security/agent-sec-core/qoder-skill-ledger-demo/)
· [Skill Ledger guide](https://agentic-os.sh/docs/user-guide/agent-security/agent-sec-core/skill-ledger/)

<table align="center" cellpadding="0" cellspacing="0">
  <tr>
    <td>
      <video
        controls
        muted
        preload="metadata"
        src="https://github.com/user-attachments/assets/aad6e296-7c5a-4a81-be2e-ea4f49e43637"
      ></video>
    </td>
  </tr>
</table>

[Choose a runtime or security starting point →](https://agentic-os.sh/docs/quickstart/)
· [Start with ANOLISA CLI](https://agentic-os.sh/docs/user-guide/user-entrypoint/anolisa-cli/)

## Install

ANOLISA CLI is the common installation entry point. cosh-ng is installed in
system mode; Token-less and other capabilities can be added independently.

```bash
curl -fsSL https://get.agentic-os.sh | bash

sudo anolisa --install-mode system install cosh-ng
anolisa install tokenless
```

Run `cosh` to enter the AI-native terminal. Token-less can also optimize tool
calls from an existing Agent without changing its framework.

[Read the Quick Start →](https://agentic-os.sh/docs/quickstart/)

## Documentation

[Quick Start](https://agentic-os.sh/docs/quickstart/) ·
[Installation](https://agentic-os.sh/docs/user-guide/installation/) ·
[User Guide](https://agentic-os.sh/docs/user-guide/) ·
[Troubleshooting](https://agentic-os.sh/docs/user-guide/troubleshooting/) ·
[Build from Source](https://agentic-os.sh/docs/building/) ·
[Changelog](https://agentic-os.sh/changelog/)

## Community

<div align="center">

<img src="docs/images/readme/dingtalk-qr.png" alt="ANOLISA DingTalk community QR code" width="180"/>

Scan with DingTalk to join the ANOLISA community.

</div>

- [Open an issue](https://github.com/alibaba/anolisa/issues) for bugs and
  feature requests.
- Read [CONTRIBUTING.md](https://github.com/alibaba/anolisa/blob/main/CONTRIBUTING.md)
  before submitting a pull request.
- Report vulnerabilities through the
  [Security Policy](https://github.com/alibaba/anolisa/blob/main/SECURITY.md).

## License

ANOLISA is released under the
[Apache License 2.0](https://github.com/alibaba/anolisa/blob/main/LICENSE).