Craft Agents: An Agent-Native Desktop App for LLM Interaction
This repository profile is provided by osrepos.com, an open source repository discovery platform.

Summary
Craft Agents is an open-source desktop application designed for effective interaction with large language model (LLM) agents. It enables intuitive multitasking, seamless connection to any API or service, and a document-centric workflow within a fluid user interface. Built with agent-native software principles, it offers high customizability and supports multiple LLM providers.
Repository Information
Topics
Click on any tag to explore related repositories
Use at your own risk
OSRepos shares public repositories for knowledge and discovery only. Any installation, execution, configuration, or use of code from these repositories is the user's own responsibility. Always review the repository, source code, dependencies, licenses, and security implications before running or installing anything. OSRepos is not responsible for issues, damages, or losses resulting from third-party repositories.
Introduction
Craft Agents is an innovative open-source desktop application built to enhance how developers and users interact with powerful AI agents. Developed by craft.do, it provides a beautiful and fluid UI that facilitates intuitive multitasking, effortless connection to various APIs and services, and a more document-centric workflow compared to traditional code-centric approaches. This project leverages the Claude Agent SDK and Pi SDK, embodying agent-native software principles and offering extensive customization.
Installation
Getting started with Craft Agents is straightforward, with both one-line installation scripts and the option to build from source.
One-Line Install (Recommended)
For macOS / Linux:
curl -fsSL https://thecraftagents.com/install-app.sh | bash
For Windows (PowerShell):
irm https://thecraftagents.com/install-app.ps1 | iex
Build from Source
git clone https://github.com/lukilabs/craft-agents-oss.git
cd craft-agents-oss
bun install
bun run electron:start
Examples
Craft Agents simplifies complex interactions, making many tasks surprisingly easy. Here are some examples of what "just works":
- Connecting to Services: Simply tell the agent, "add Linear as a source." It automatically discovers public APIs, reads documentation, sets up credentials, and configures everything without manual config files or setup wizards.
- Custom APIs and Local MCPs: Paste an OpenAPI spec, endpoint URLs, or even screenshots of documentation, and the agent will figure it out. Local MCP servers and direct database connections are fully supported.
- Skill Management: Import your Claude Code skills with a single command, or describe a new skill, and the agent handles its creation. Changes are instant, with no restarts required.
- CLI Client: The project also includes a powerful CLI client for scripting, CI/CD pipelines, or command-line enthusiasts. You can run self-contained prompts, manage sessions, and interact with the server.
# Self-contained run (spawns its own server) craft-cli run "Summarize the README" craft-cli run --workspace-dir ./my-project --source github "List open PRs"
Why Use It
Craft Agents addresses the need for a more intuitive and powerful way to work with AI agents. Its key advantages include:
- Agent-Native Workflow: Designed from the ground up for agents, it allows you to describe what you want, and the agent figures out how to achieve it, optimizing token usage.
- Intuitive Multitasking & UI: A beautiful and fluid desktop UI provides session management, status workflows, and a document-centric approach that enhances productivity.
- Extensive Connectivity: Connects to virtually any API or service, including popular platforms like Linear, Gmail, Slack, and custom APIs, without complex configurations. It also supports local filesystems and Git repositories.
- Multi-LLM Provider Support: Seamlessly integrate with various LLM providers such as Anthropic, Google AI Studio, ChatGPT Plus, GitHub Copilot, and even self-hosted models via custom endpoints.
- High Customizability & Automation: Create specialized "skills" for agents, define permission modes, and set up event-driven automations to streamline repetitive tasks, such as daily briefings or urgent label notifications.
- Remote Server Capability: Run Craft Agents as a headless server on a remote machine, enabling long-running sessions, multi-machine access, and offloading compute-heavy tasks.
Links
- GitHub Repository: https://github.com/craft-ai-agents/craft-agents-oss
- Demo Video: https://www.youtube.com/watch?v=xQouiAIilvU
- Official Website: https://thecraftagents.com/
Related repositories
Similar repositories that may be relevant next.

Education Agent Skills: 165 AI Skills for Evidence-Based Teaching
August 24, 2026
The Education Agent Skills library offers 165 evidence-grounded AI skills designed for teachers, school leaders, and EdTech builders. It provides a robust foundation for AI in education, focusing on pedagogy, learning science, curriculum, and assessment. Compatible with AI agents like Claude, Codex, and Hermes, this open-source resource aims to enhance teaching quality and student outcomes.

agent-memory: A Standalone Memory System for AI Coding Agents
August 23, 2026
agent-memory is a standalone memory system designed for AI coding agents, addressing the common problem of agents "forgetting" context between sessions. It provides a structured, four-layer pipeline for capturing, consolidating, and disclosing knowledge. This system ensures agents always start with relevant context, eliminating the need for manual keyword-dependent retrieval.

Codex MCP Server: Bridge Claude Code with OpenAI Codex CLI
August 23, 2026
Codex MCP Server is an essential tool that acts as a bridge between Claude Code and OpenAI's Codex CLI. It allows developers to integrate powerful AI-driven code analysis, generation, and review capabilities directly into their editor. This server wrapper enables Claude Code to fully leverage Codex's advanced AI functionalities.

AgentBridge: Local Bridge for Claude Code and Codex Collaboration
August 22, 2026
AgentBridge is an innovative local tool designed to facilitate real-time, bidirectional collaboration between Claude Code and Codex AI agents. It enables seamless cross-review, task splitting, and quota relay, allowing these powerful models to work together efficiently within a single session without manual intervention. Notably, the tool itself was largely built by Claude Code and Codex collaborating through AgentBridge.
Source repository
Open the original repository on GitHub.