craft-agents-oss
SHA-256Craft Agents is an open-source desktop app for working with AI agents, featuring intuitive multitasking, universal API/MCP connectivity, and document-centric workflow.
Smart Download
Download Download Version
v0.10.5 · 277 MB
A graphical, open-source desktop app for AI agents with multi-session inbox, instant API/MCP connections, and no restarts.
Core Features
- Multi-session inbox with status workflow, flagging, and archiving
- Connect MCP servers, REST APIs (Google, Slack), local filesystems, and more
- Three permission modes (Explore, Ask to Edit, Auto) switchable with Shift+Tab
- Background tasks with progress tracking
- Multi-LLM support: Anthropic, Google AI Studio, ChatGPT Plus, GitHub Copilot
What It Can't Do
- •Requires valid API keys or subscriptions (e.g., Anthropic, OpenAI). Internet connection needed for initial setup. Remote server mode requires TLS configuration (use a reverse proxy for production). Local skills and sources are stored in the workspace directory; back up before uninstalling.
Use Cases
- Automate repetitive tasks with AI agents (e.g., email triage, report generation)
- Connect project management tools (Linear, Notion) and communication platforms (Slack) via natural language
- Create custom automation workflows triggered by label changes, schedules, or tool usage
Detailed Introduction
Craft Agents is an open-source desktop application designed to streamline collaboration with AI agents. Built on Claude Agent SDK and Pi SDK, it offers a graphical interface (non-CLI) for managing multiple agent sessions, connecting to any API or MCP server, and organizing work in a document-centric way. Unlike CLI-based tools like Claude Code, it provides an inbox, permission modes, background tasks, and a dynamic status system. It supports multiple LLM providers (Anthropic, Google AI Studio, ChatGPT Plus, GitHub Copilot) and allows instant skill/source import without restarts. The project is highly customizable, built with agent-native principles, and offers a remote headless server mode for running long-lived sessions on a VPS.
Tags
Getting Started
Download installer
Click the button above to download the installer for your system
Install the software
Install the appropriate package for your distro (dpkg / rpm / AppImage)
On macOS/Linux: run curl -fsSL https://agents.craft.do/install-app.sh | bash
On Windows PowerShell: run irm https://agents.craft.do/install-app.ps1 | iex
Launch the app and choose an API connection (Anthropic, Google AI Studio, etc.)
- On macOS/Linux: run curl -fsSL https://agents.craft.do/install-app.sh | bash
- On Windows PowerShell: run irm https://agents.craft.do/install-app.ps1 | iex
- Launch the app and choose an API connection (Anthropic, Google AI Studio, etc.)
SHA-256 checksum verified
Checksum extracted from GitHub official Release page
SHA256 Checksum
2595ea80e3afac2fb418f568adba9f6fdad535e7a46bc91b3af7a9053ff7796cThis checksum is extracted from the GitHub Release page. Verify file integrity after download.
All SHA-256 checksums on this platform are extracted from the project's official GitHub Release page, without any modification. You can independently verify them on the GitHub Releases page.
Open Source Transparency
View GitHub SourceUninstall Info
Remove the application (drag to Trash on macOS, uninstall via Control Panel on Windows) and manually delete the data directory ~/.craft-agent if needed.
No Extra Dependencies
Ready to use after download. No additional runtime required.
Similar Projects
AnythingLLM
Chat with your docs, use AI agents, multi-user support, runs locally with zero setup.
goose
Goose is a general-purpose AI agent that runs on your machine. Use it for coding, research, writing, automation, and more. It offers a native desktop app (macOS, Linux, Windows), a full CLI, and an API. Works with 15+ providers (Anthropic, OpenAI, Google, Ollama, etc.) and connects to 70+ extensions via the Model Context Protocol. Built in Rust, part of the Linux Foundation.
ollama
Ollama lets you download, run, and manage large language models locally. One command, multiple platforms, endless possibilities.