Maka is a local-first AI desktop workbench enabling users to run a controllable, observable, and recoverable AI agent on their own computers. It targets end-users configuring local AI and engineers developing the platform, offering a unified environment for model connections, session management, tool permissions, file I/O, terminal execution, search, and bot integrations, thereby enhancing privacy and data control.

How It Works

Built as an Electron application, Maka integrates diverse AI model providers, including cloud APIs (OpenAI, Anthropic, Gemini, DeepSeek, Moonshot) and local models via Ollama. It utilizes the Vercel AI SDK for a robust runtime supporting streaming, tool invocation, usage tracking, and session recovery. Core functionalities like file operations and shell commands are managed by a permission engine for user oversight. Data, including credentials (encrypted), is stored locally within the Electron user data directory, reinforcing its local-first, privacy-centric design.

Quick Start & Requirements

Installation requires cloning the repo, running npm install, and then npm run dev. The project uses npm workspaces. Prerequisites include Node.js and npm. Manual Electron binary installation is needed if ELECTRON_SKIP_BINARY_DOWNLOAD=1 is set. Optional environment variables like OPENAI_API_KEY can bootstrap initial model connections. Detailed documentation is available in the docs/ directory.

Highlighted Details

• Broad Model Support: Integrates major cloud AI providers and local Ollama instances, plus custom compatible endpoints.
• Local Tooling: Provides Read, Write, Edit, Bash, Glob, and Grep tools governed by a permission system.
• Privacy-Focused Architecture: Emphasizes local data storage and encrypted credential management.
• Extensible Framework: Supports integration with messaging bots (Telegram, Discord, WeChat) and network search (Tavily).

Maintenance & Community

The repository is under active development. It features comprehensive testing plans for UI quality, visual regression, and full product QA, detailed in its documentation. Contribution guidelines recommend running type checks, builds, and focused tests.

Licensing & Compatibility

The README does not specify a software license. This omission prevents determining compatibility for commercial use or closed-source linking without further clarification.

Limitations & Caveats

As the project is actively developing, expect potential feature changes. Setup requires a Node.js environment and may involve manual Electron binary steps. The most significant limitation for adoption is the lack of a declared software license.