Doka (Doka AI) is a desktop AI agent whose official positioning is "not a prompter, but an agent"—chat tools only answer questions, while Doka directly gets work done on your computer.
Purpose
Doka is designed to let AI perform actual tasks on a local computer, rather than merely giving suggestions or text replies. It can view your files, run commands, and push tasks through to completion. The tool is suitable for scenarios that require AI to directly operate the file system and terminal, such as code modifications, batch file renaming, and report organization.
Main capabilities
- Desktop operation: Doka is installed on your computer, rather than existing only as a web service.
- File and terminal operations: able to view files, run commands, and execute multi-step tasks.
- Model selection: users can choose to use cloud models or local models themselves.
- Local operation and privacy: files and context do not leave your device unless you actively want them to.
- Extensibility: supports MCP tools and custom scenario scripts.
- Task execution rather than just answering: can modify code, rename files, generate reports, and advance step by step.
- Adapts to different work scenarios: the official website gives examples such as builders, lawyers, and analysts; Doka adapts to workflows rather than requiring users to change their habits.
How to use
According to the official website information, users can download Doka for free for personal use, with support for Windows, macOS, and Linux. After installation, Doka runs in the desktop environment, can connect to cloud or local models, and can be configured through MCP tools and custom scenarios. Specific installation steps and configuration details are not yet expanded in the provided materials.
Applicable scenarios
- Software development: modify code, handle project files, run terminal commands.
- File management: batch renaming and organizing files.
- Reports and documents: automatically collect information and generate reports.
- Multi-step tasks: break down complex processes and execute them step by step.
- Privacy-sensitive work: models can run locally, and files do not need to be uploaded to the cloud.
Limitations and information not yet provided
- Pricing model: the official website only states that personal use is free to download; pricing for other versions or commercial use is not yet provided.
- Specific model list: the materials mention that cloud or local models can be selected, but do not list the specific supported model names.
- Platform details: although Windows, macOS, and Linux are supported, the minimum system requirements for each platform are not yet provided.
- Installation and configuration steps: the official website does not provide a detailed installation tutorial in the provided content.
- Supported MCP The specific formats and examples of tools and custom scenarios are not yet provided.
The official website also has a technical articles section, with recent content including ffmpeg Windows installation, Godot 4 download and configuration, and troubleshooting Python ModuleNotFoundError, which can serve as auxiliary references.
