Verified project record

apache/maka

Apache Maka is an agent workspace available as a desktop app, TUI, and CLI that durably records model messages, tool calls, permission decisions, and task execution in an append-only runtime event log. Users supply their own model connections (API, local model, or supported account) rather than a bundled shared model account.

Project overview

Its append-only RuntimeEvent log is not just a record: it drives the UI, the next prompt, and crash recovery, and the project also ships an evaluation framework with experiment cells and adapters.

Project type: AI Agent · AI Coding · Evaluation & Observability
Use cases: Coding & Development · Automation
Deployment: Refer to project documentation
License: Apache-2.0

Best for

Developers and AI engineers who want an agent workspace that durably records model messages, tool calls, permission decisions, and task execution and who can configure their own model connection.
Users who want sessions, settings, and run records to stay local while supplying their own model.

Key capabilities

Provides a terminal UI and non-interactive CLI for starting turns and running repository tasks.
Records model messages, tool calls, permission decisions, and termination as append-only RuntimeEvents used as the basis for UI, prompts, and crash recovery.
Lets users add an API, local-model, or supported account connection and choose a default model instead of bundling a shared model account.
Requires tools that write files or run a shell to pass the sandbox boundary first.
Ships experiment cells, attempts, results, and executor or subject adapters for evaluating Maka subjects.
Supports explicit interrupted-turn resume in Desktop and through the CLI/TUI resume command.
Includes agent skills in the repository's skills directory.

Limitations and risks

Maka has not made an Apache release yet; packages distributed elsewhere are convenience artifacts.
Windows and Linux are identified as preview platforms in the repository overview.
API keys and similar secrets are stored in a local plaintext credential file readable by the OS account.
Data formats, CLI commands, and experimental capabilities may still change while the project is under active development.
Older JSONL transcripts and Electron safeStorage credential files are not imported during workspace upgrades.

Getting started

git clone https://github.com/apache/maka.git, cd maka, npm ci, npm run dev. Requirements: Node.js 22.19 or newer, npm, Git, and ripgrep. On first launch, open Settings → Models, add an API, local-model, or supported account connection, test it, choose a default model, and start a task.

Evidence and sources

README: Apache Maka (Incubating) is a high-performance agent workspace that keeps a complete record of everything it did.
README: Every model message, tool call, permission decision and termination is an append-only RuntimeEvent. The UI, the next prompt and crash recovery are projections of that log, never t…
README: Sessions, settings and run records stay local. You bring the model: a cloud API, a local model or a compatible gateway.
README: apps/desktop/ Electron main / preload / React renderer packages/core/ Pure contracts for Sessions, Events, Permissions, and Connections packages/storage/ SQLite operational state,…
README: Apache Releases: Maka has not made an Apache release yet.

Source updated at: 2026-09-28T03:23:27Z

Open original GitHub project