Run a local coding agent with Ollama + Qwen 3 using OpenPilot

Description and use

OpenPilot enables running a local coding agent with Qwen 3 models via Ollama, allowing safe file edits and shell command execution within a chosen project folder.

Problem

Need an open‑source desktop coding agent that can run on a local machine with any model, including Qwen 3, and allows safe file edits and shell commands.

Outcome

OpenPilot successfully runs locally with Qwen 3 models via Ollama, allowing file edits and shell command execution within a project folder, with user approval for sensitive actions.

Workflow

• Pull a Qwen 3 model with "ollama pull qwen3:8b" or "ollama pull qwen3:27b"
• Ensure Ollama is running (ollama serve or open the app) to expose an OpenAI‑compatible API at http://localhost:11434/v1
• Install OpenPilot: download the Windows installer or macOS .dmg, or build from source with Node.js 20+ and npm install
• Launch OpenPilot and connect the model: set Name to the exact model name from "ollama list", Base URL to http://localhost:11434/v1, and any placeholder API key
• Select a workspace folder (preferably a throwaway or git‑tracked project)
• (Optional) enable TinyFish for web search if online access is desired
• Give the agent a concrete task, e.g., "Read README.md and add a 'Running tests' section based on package.json scripts."
• Approve or deny any sensitive actions presented by the agent

Evidence

I wanted something like Cursor that was open source and let me use whatever model I wanted, including ones running on my own machine.

Source

I've tested Qwen 3 8B and Qwen 3 27B this way (through Ollama and LM Studio) and both worked well for me.

Source

Make sure Ollama is running (ollama serve, or just open the Ollama app). It serves an OpenAI-compatible API at http://localhost:11434/v1.

Source