Repolode
Submit
apache

apache/maka

Apache Maka (Incubating) is an agent workspace for running AI coding tasks locally, with a full append-only record of every model message, tool call and permission decision.

Open-source alternative to:Claude CodeCursor

View on GitHub
Stars
5,846
Forks
557
Open issues
389
Last commit
8 hours ago

What is it?

Apache Maka is an agent harness and workspace for completing tasks with a language model. It is available as a desktop app, a terminal UI and a non-interactive CLI, and all three are clients of a single Runtime Host that executes agent turns.

Every model message, tool call, permission decision and termination is recorded as an append-only RuntimeEvent. The UI, the next prompt and crash recovery are built from that log, so old tool output can be dropped from the prompt while remaining in the record. Sessions, settings and run records are stored locally, and the user supplies the model through a cloud API, a local model or a compatible gateway.

The project includes an evaluation package for running experiments on the same model with an official verifier and publishing per-task results. Desktop builds are documented for macOS, with Windows and Linux listed as previews. The project is under Apache Incubator and its data formats and CLI commands may still change.

Key features

  • 01Append-only event log that serves as the source of truth for the UI, prompts and recovery.
  • 02Runs as a desktop app, a TUI, or a non-interactive CLI sharing one Runtime Host.
  • 03Keeps sessions, settings and run records on the local machine.
  • 04Supports cloud APIs, local models and compatible gateways.
  • 05Includes an evaluation package for benchmarking runs with per-task records.
  • 06Permission prompts for tool calls such as shell commands and file edits.

Use cases

  • Running coding agents on a local repository while keeping a full audit trail of actions.
  • Comparing agent harnesses on the same model with reproducible per-task results.
  • Using a local or self-hosted model through an agent workspace without sending session data to a hosted service.
  • Driving agent tasks from a terminal or from scripts via the non-interactive CLI.

Topics

#agent-runtime#ai-agent#llm#tool-use#electron#typescript#local-first#event-sourcing#ai#apache

FAQ

What is maka?
Apache Maka (Incubating) is an agent workspace for running AI coding tasks locally, with a full append-only record of every model message, tool call and permission decision. Apache Maka is an agent harness and workspace for completing tasks with a language model. It is available as a desktop app, a terminal UI and a non-interactive CLI, and all three are clients of a single Runtime Host that executes agent turns.
Is maka free to use? What license does it have?
maka is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is maka still maintained?
Actively maintained. The last commit to maka was on Oct 18, 2026.
What can I use maka for?
Running coding agents on a local repository while keeping a full audit trail of actions. Comparing agent harnesses on the same model with reproducible per-task results. Using a local or self-hosted model through an agent workspace without sending session data to a hosted service. Driving agent tasks from a terminal or from scripts via the non-interactive CLI.
Where can I get maka?
The source code and installation instructions are on GitHub: https://github.com/apache/maka

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects