SkillFoxx

ANOLISA CLI

ANOLISA is a system stack for AI agents: output compression, memory, workspace snapshots, command checks and token accounting

  • 663 ★ on the repository
  • Last commit
  • Apache-2.0
CLI

Broad access

Why this level

  • Changes how the agent environment runs: commands pass through AW and components work with the kernel and system parameters
  • Workspace snapshots, memory and execution records hold file contents and command output
  • The `ktuner` component can apply and roll back Linux kernel parameters
All reasons and checks

agentic-os-org/anolisaAuthor:Agentic OS

Install

Manual install

aw config init --prefix "$AW_PREFIX" --config "$AW_HOME/anolisa.yaml" --state-dir "$AW_HOME/anolisa-state" --qoder "$(command -v qodercli)" --provider sec-core+tokenless --socket "$AW_SEC_SOCKET"

Creates a config with command checks and output compression. The agent starts with `aw run --config "$AW_HOME/anolisa.yaml" --agent qoder`.

Access and checks

How we review

Broad accessThe tool can change data, run commands or handle secrets.

  • Changes how the agent environment runs: commands pass through AW and components work with the kernel and system parameters
  • Workspace snapshots, memory and execution records hold file contents and command output
  • The `ktuner` component can apply and roll back Linux kernel parameters

This is third-party code. Review the repository files before installing.

What it does

ANOLISA is a set of system components for AI agent workloads on top of an existing OS and kernel. The agent stays the same and the stack adds to its environment. Tokenless compresses tool output and simplifies schemas. Agent Memory keeps memory across sessions, ws-ckpt takes workspace snapshots so files can be compared and rolled back, and Agent Security Core checks commands before they run. AgentSight shows model calls, trajectories and token usage. For people there is the `cosh` terminal, which keeps a familiar shell and calls tools and skills through natural-language phrases. OS Skills is a handbook for agents on administration, monitoring, security and DevOps. SkillFS lays skills out as stable filesystem views, and `ktuner` inspects Linux kernel parameters. Components reach an agent through the AW layer: adapters and plugins enable command checks and output optimization without changing the agent's interface. The documented example currently uses an adapter for `Qoder`, and the commands `aw config init`, `aw validate`, `aw run`, `aw status` and `aw inspect` drive the run and inspect execution records. The ANOLISA CLI manages component installation.

Who it is for. Platform teams running agents on Linux servers and in sandboxes that want shared snapshots, command checks and token accounting.

Good fit when

  • An agent often reads long tool output that needs compressing without changing the agent itself
  • After an agent run you need to diff workspace files and roll back extras
  • Agent commands must pass a safety check before they run on a shared server
  • A failure review needs the trajectory: model calls, tool calls and token usage
  • A sandbox image is built with agent components preinstalled

Not a fit when

  • You run a single local agent on a laptop: the stack targets server workloads on Linux and macOS
  • You need ready adapters for Claude Code or Cursor: the README documents the `Qoder` path and leaves the rest to the open architecture
  • You want a single skill or MCP server, while ANOLISA installs as a set of components

Example request

Run the agent through aw with command checks and output compression, and after the task show the execution records with aw inspect

Limitations

The stack is a preview: the README example needs Linux x86_64 with glibc 2.39 or newer, `Qoder` CLI 1.1.64 or later signed in, and a running security backend. A ready path is documented for `Qoder`, and other agents need adapters through the open AW architecture. Each component supports its own set of platforms, listed in the installation guide. The `copilot-shell` component has moved to a deprecated folder. Ready images and cloud services are listed in the image catalog.

How to disable. Remove components through the ANOLISA CLI or the package manager you used. Stop the AW service with `aw stop --config "$AW_HOME/anolisa.yaml"` and the security backend the way you started it.

FAQ

What is ANOLISA?

A multi-component system stack for AI agents: it adds output compression, memory, workspace snapshots, command checks and observability without replacing the agent.

How do I attach ANOLISA to an existing agent?

Through the AW layer: an adapter or plugin enables checks and optimization in the agent's environment. The README gives a `Qoder` example with `aw config init` and `aw run`.

What does Tokenless do?

It compresses tool output, simplifies tool schemas and rewrites commands so the agent spends fewer tokens. In the example it compresses text results after a successful call.

How do I roll back an agent's changes in the workspace?

The Workspace Checkpoint component (ws-ckpt) takes workspace snapshots, so you can compare files and restore the version you want.

Which platforms are supported?

The page lists Linux and macOS, but each component has its own list in the installation guide. The AW example needs Linux x86_64 and glibc 2.39.

Discussion

No questions or reviews yet.

CLI tools

A command-line SSH tool for agents: remote execution, file upload and download, and command allow and block lists.

CLIBroad access88

A Docker sandbox for coding agents: a Debian image with Copilot CLI, Codex and Pi, and SSH, RDP and Docker enabled by flags.

CLIExtended access154

Allox CLI creates an isolated agent environment, runs shell, browser and Jupyter, saves checkpoints and restores work.

CLIBroad access163

CLI that runs commands and AI agents in isolated microVMs that boot in under 125 ms and keeps API keys out of the sandbox itself.

CLIBroad access62

A skill and Python library: four agent decision-quality indicators, a repeated-write guard and a circuit breaker for agent chains

SkillExtended access309
Official

A platform on your own server that gives each agent its own Linux box with SSH and MCP tools for commands and files

MCP serverBroad access318

README in short

The README defines ANOLISA as a system stack for agent workloads and offers three paths: a ready image, installation on an existing Linux system, or attaching an agent you already use. It then explains why agents need a stack, shows video examples, tables of components by layer and the open AW architecture with a provider protocol. The page closes with a setup and run example and notes on development and documentation.

README badge

Report a problem