ANOLISA

<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"

Agentic Nexus Operating Layer & Interface System Architecture

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 · Website · Quick Start · User Guide ·

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?

Components

Agent entry Context efficiency Runtime &amp; security cosh-ngShell copilot Token-lessTool-output compression ws-ckptCheckpoint and rollback OS SkillsSystem and DevOps expertise AgentSightTrace and Token visibility SkillFSFocused Skill views ktunerKernel tuning Agent MemoryCross-session memory Agent Sec CoreSandbox and verification AWUnified Agent policy configuration

Blaze has moved to its standalone repository, blaze-rs.

What it solves

Let the Agent work directly in the terminal

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.

See where Tokens go and cut waste before it reaches the model

removes redundancy from tool schemas and responses before they reach the model. reuses useful context across sessions. SkillFS keeps the current Skill view focused and makes other Skills discoverable when needed. 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.

<video controls muted preload="metadata" src="https://github.com/user-attachments/assets/ed6952c6-19da-44c7-ae35-85d753bdccf9"

Try Token-less with Claude Code in 3 minutes

Install Token-less and connect it to Claude Code:

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:

tokenless stats summary
tokenless stats list --limit 5

· Read the user manual

<video controls muted src="https://github.com/user-attachments/assets/b372ae72-44fa-492f-9feb-e6cd137b631a"

In one observed coding task, Token-less saved 317K Tokens (40.5%), based on AgentSight measurements. Results vary by workload.

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 > 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 explains how to estimate the effect for a given workload.

Give every Agent execution a boundary and a way back

ANOLISA is building out the Agent execution environment: isolates risky operations, and 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.

· Skill Ledger guide

<video controls muted preload="metadata" src="https://github.com/user-attachments/assets/aad6e296-7c5a-4a81-be2e-ea4f49e43637"

· Start with 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.

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.

Standards & Compliance

Standard / Framework · Status and details AARM · Aligned · Registry entry: AgentSecCore (ANOLISA) OWASP Agentic Top 10 (2026) · All 10 ASI risk categories mapped to security controls

Documentation

Quick Start · Installation · User Guide · Troubleshooting · Build from Source ·

AW: Architecture · User guide · Build and package ·

Repository layout

Components remain independently buildable in one repository. Directory placement separates runtime capabilities, distribution tooling, evaluation, and retained legacy code; it does not rename installed commands, packages, or release tags.

Directory · Contents src/ · Runtime components, including AW, cosh-ng, observability, security, memory, and sandbox capabilities. Each component keeps its own build and test entry points. providers/ · AW Provider collaboration: independent implementation workspaces, examples and contract tests. distribution/ · Distribution tooling. anolisa/ owns the installation CLI, component manifests, and packaging resources. benchmark/ · Cross-component evaluation runners: ClawEval, SWE, and Terminal Bench. Component-specific benchmarks stay with their components. deprecated/ · Retained legacy implementations and their archived documentation. scripts/ · Repository-wide build, packaging, and validation scripts. tests/ · Repository-wide regression checks and test orchestration. Component tests remain local to each component. docker/ · Container image definitions and build resources. docs/ · User and developer guides, source-build instructions, and shared documentation assets. specs/ · Contribution standards, component onboarding, and documentation rules. website/ · Public website source and generators. Generated output is not a second documentation source. .github/ · CI, release workflows, component routing metadata, and ownership.

Before changing files, start with AGENTS.md for repository rules, component locations, and build/test commands, then read the component's scoped AGENTS.md when present. Use CONTRIBUTING.md for the contribution workflow. Source-path changes must be reflected in .github/components.json and its build, release, ownership, and documentation consumers without changing the component's public identity.

Community

Scan with DingTalk to join the ANOLISA community.

• Open an issue for bugs and

feature requests.

• Read CONTRIBUTING.md

before submitting a pull request.

• Report vulnerabilities through the

Security Policy.

License

ANOLISA is released under the Apache License 2.0.