Name: OpenAgentd
Availability: InStock
Author: lthoangg

What is OpenAgentd?

OpenAgentd is a desktop program that runs a coding agent locally on your computer. You point it at a project folder and describe a task; the agent then plans, reads files, makes edits, executes tests, and streams every action — tool invocations with their arguments and timing, file diffs, and token usage — into a single transcript. You can intervene during a turn, review changes before they are committed, and choose from over 19 model providers including major cloud services, GitHub Copilot, OpenAI Codex, or local models via Ollama. All sessions, telemetry, and configuration remain on your device unless you enable network access for mobile clients. The backend is a single Rust binary (roughly 43 MB, about 20 MB idle) wrapped in a Tauri interface. Installers exist for macOS, Windows, and Linux, plus a Homebrew cask. Mobile apps for iOS and Android link to the desktop backend over the local network, and a headless CLI server mode is available for browser or remote access. Capabilities include a read-only planning mode with approval workflow, multi-agent teams, live web preview, scheduled prompts, a plugin system, Model Context Protocol integration, path denylists for secrets, cross-workspace messaging, persistent memory, cost and latency tracking, and Git worktree support. The codebase is licensed under Apache 2.0.

Should you self-host OpenAgentd?

OpenAgentd scores 51/100 on selfhost.directory (how scoring works) and ranks #33 of 120 in AI Assistant. It has 213 GitHub stars.

It shipped 64 stable releases in the past year. The latest, v3.34.0, came out on 27 October 2026.

Setup is easy: most people get it running in one sitting. It is Apache-2.0-licensed and written mostly in Python.

Its closest alternative on selfhost.directory is hermes-agent (89/100, 257 957 stars). It scores higher overall, so compare the two before you pick. See OpenAgentd vs hermes-agent side by side.

FAQ

Can I keep secrets away from the agent?

Yes. A path denylist blocks files such as .env from the agent's file tools with glob patterns, and shell commands start without the API keys and other secrets OpenAgentd holds.

Can I use it from my phone?

Yes. The iOS and Android apps connect to OpenAgentd on your computer. Turn on Network access in Settings and enter the address and access key, or run the CLI server with an access key.

What happens when I hit a usage limit or lose the network?

When a provider runs out of quota, the agent shows a countdown, waits, and carries on. If the network drops, it retries every few seconds until the connection is back.

What tools does the agent have?

It reads, searches, and edits files, runs shell commands, navigates code and reads diagnostics through language servers. It can also search and fetch web pages, generate images and videos, keep a task list, schedule prompts, start helper agents, and message agents in other workspaces.

Is there a CLI?

Yes. Install the CLI server to run agentd server start and connect a browser or another device. agentd run runs an agent turn in the current project.

What does it cost?

OpenAgentd is free and open source under Apache 2.0. You pay your model provider for what you use, or nothing with local models.

Which models can I use?

Any of 19+ providers including Anthropic, OpenAI, Gemini, OpenRouter, AWS Bedrock, GitHub Copilot, and Ollama. Use your own keys or sign in with GitHub Copilot or OpenAI Codex subscription.

Does my code leave my machine?

Your prompts and the files the agent reads go to the model provider you choose. Sessions, telemetry, and settings stay on your computer. With a local model through Ollama, nothing leaves it.