Maka: A High-Performance Agent Workspace for AI Tasks

Introduction

Apache Maka (Incubating) is a high-performance agent workspace that keeps a complete record of everything it did. As an Apache Incubating project, Maka is built to help AI agents complete tasks efficiently and transparently. It emphasizes measurable performance, ensuring that every run is benchmarked and fully recorded.

Maka operates on a few core principles: its performance is measured, not just claimed, with per-task results published. The entire runtime is based on an append-only log of RuntimeEvents, which drives the UI, future prompts, and crash recovery. It supports a local-first approach, keeping sessions, settings, and run records on your machine, allowing you to bring your own model, whether it's a cloud API, a local model, or a compatible gateway. Furthermore, Maka provides a single Runtime Host that serves as the execution authority for its Desktop, TUI, CLI, and evaluation clients.

Why Use and Key Benefits

Maka offers several compelling advantages for developers and researchers working with AI agents:

• Measured, Not Claimed Performance: Unlike many agent harnesses, Maka is rigorously benchmarked against others using the same model and official verifier. The per-task results are transparently published, providing concrete evidence of its efficiency and effectiveness.
• The Log is the Runtime: Every significant action, including model messages, tool calls, permission decisions, and terminations, is recorded as an append-only RuntimeEvent. This robust logging mechanism ensures that the UI, subsequent prompts, and crash recovery are all projections of this immutable log, enhancing reliability and auditability.
• Your Machine, Your Model: Maka is designed with a local-first philosophy. All sessions, settings, and run records remain on your local machine, ensuring data privacy and control. It offers flexibility in model integration, allowing users to connect to various models, including cloud APIs, local models, or compatible gateways.
• One Runtime Host: The project unifies its execution authority through a single Runtime Host. This means that whether you are using the Desktop application, the Terminal User Interface (TUI), the Command Line Interface (CLI), or the evaluation framework, they all leverage the same core execution logic, ensuring consistent and predictable agent behavior across different interfaces.

Installation

To get started with Apache Maka, you can build it directly from source. Here are the steps to set up and run the Desktop application:

Requirements:

• Node.js 22.19 or newer (CI uses Node.js 24)
• npm (the lockfile and scripts use npm; the current packageManager is npm 11)
• Git
• ripgrep, used by Runtime's Grep tool

Start Desktop:

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

npm run dev starts the Desktop development environment with Hot Module Replacement (HMR). To build every workspace before starting Electron, use npm run dev:full.

First Run Configuration:

Maka does not bundle a shared model account. On first launch, you will need to:

1. Open Settings ? Models.
2. Add an API, local-model, or supported account connection.
3. Test it and choose a default model.
4. Return to the workspace and start a task.

Examples

Maka provides powerful command-line tools for interacting with agents. After building the workspaces with npm run build, you can use the development CLI:

Start the TUI:

npm run cli:dev

Run a specific task:

npm run cli:dev -- run "Summarize this repository and identify its most important risk"

Run a task with graph visualization:

npm run cli:dev -- run --graph "Implement two independent slices, integrate them, then review the result"

Links

• Website: https://maka.apache.org/en/
• Documentation: https://maka.apache.org/en/docs/
• GitHub Repository: https://github.com/apache/maka