Apache Maka CLI

Apache Maka (Incubating)

An agent workspace with a full log of every step: model messages, tool calls and permission decisions as one append-only event chain

  • 5,962 ★ on the repository
  • Last commit
  • Apache-2.0
CLI

Extended access

Why this level

  • Runs commands and tools with permissions on your machine
  • No signed official release yet, only building from source
All reasons and checks

apache/maka

Install

Manual install

git clone https://github.com/apache/maka.git
cd maka
npm ci
npm run dev

There is no official Apache release yet; this is a development build from source.

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • Runs commands and tools with permissions on your machine
  • No signed official release yet, only building from source

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

What it does

Maka is an Apache Software Foundation agent harness where the runtime log is the single source of truth: every model message, tool call, permission decision and turn termination is recorded as a RuntimeEvent, and the UI, the next prompt and crash recovery are all built on top of that log rather than stored separately. Desktop, the TUI, the CLI and the eval module all run through the same Runtime Host, and comparisons against other harnesses on the same models are published together with the per-task logs. The model and session data stay local; you can plug in a cloud API, a local model or a compatible gateway.

Who it is for. For developers and researchers who care about agent transparency and want to verify exactly what the agent did and why.

Good fit when

  • You need a full, auditable log of every agent step for review or debugging
  • You want to compare harnesses on the same model with a transparent benchmark methodology
  • It matters that sessions and settings stay only on your machine

Not a fit when

  • You need a ready installer right now: there is no official Apache release yet, only building from source
  • A regular coding agent with no detailed runtime log is enough

Example request

Show the full log of the last turn: which tools were called and which permissions were requested

Limitations

There is no official Apache release yet, only building from source via npm, which needs Node.js 22.19 or newer and ripgrep. The project is in Incubating status.

How to disable. Remove the cloned repository and the Maka settings folders created on first run.

FAQ

Is there a ready installer?

There is no official Apache release yet, only building from source; packages distributed elsewhere are not considered an official release.

Where does session data go?

Sessions, settings and run records stay local; you bring your own model connection.

Discussion

No questions or reviews yet.

CLI tools

A self-hosted AgentOS for one workspace: plan-mode chat, private knowledge base, MCP tools, skills, memory and a sandbox

CLIBroad access631Repository stars

A local-first AI agent running on your own machine via llama.cpp that drives a browser, edits files, runs approved commands and remembers context across session

CLIBroad access3.6KRepository stars

A bridge between local CLI agents: Codex, Cursor or Claude Code delegate work to Antigravity, Grok, Kimi and seven more.

MCP serverBroad access131Repository stars

A local dashboard for running many agents across machines: memory, code graph and agent-to-agent messaging

CLIBroad access851Repository stars

A layer over coding agents that assembles a temporary specialist team per task inside Claude Code, Codex, Cursor and Gemini

PluginExtended access1.5KRepository stars

Agav is a terminal coding agent that reads a repository, edits files, runs tests and switches between seven model providers.

CLIExtended access178Repository stars

README in short

The README stresses that Maka benchmarks itself against other harnesses on the same model with the same official verifier and publishes per-task results in docs/eval. The core idea: the log is the runtime, everything else, the UI, the next prompt, crash recovery, is built on top of append-only events. Desktop, the TUI, the CLI and Eval all run through one Runtime Host. There is no official Apache release yet, so using it means building from source via npm with Node.js 22.19+ and ripgrep. The project is in Apache Incubator status.

README badge

Report a problem