Alden AI

A local-first AI coding agent for VS Code, powered by Ollama.

Your code never leaves your machine. No API keys, no subscriptions, no cloud.

What Alden Does

Alden is a full coding agent — not just a chat window. It reads your codebase, makes a plan, writes and edits files, runs terminal commands, and reports what it did. You stay in control at every step.

• Ask questions about your code and get direct answers
• Request changes and watch Alden explore, plan, and implement them
• Troubleshoot errors by pasting output — Alden diagnoses and fixes
• Multi-turn chat — Alden remembers files it has read across a conversation
• Pause and steer — stop mid-run, give a new instruction, resume
• Full chat history — every conversation is saved and reloadable

Requirements

• Ollama installed and running locally
• At least one model pulled — recommended:

  ollama pull qwen2.5:14b
  ollama pull qwen2.5-coder:14b
  

• VS Code 1.85 or later
• 8 GB RAM minimum (16 GB recommended for 14b models)

Getting Started

1. Install the extension and open VS Code
2. Start Ollama: ollama serve
3. Click the Alden AI icon in the Activity Bar (left sidebar)
4. Select your model tier in Settings, or let Alden auto-detect based on your RAM
5. Type a prompt and press Enter or click Chat

How It Works

Every request goes through up to three phases:

You can run Plan only (read-only, great for questions and analysis) or the full Plan + Act chain for code changes.

Pause, Steer, Resume

At any point during a run:

• Stop — pauses at the next checkpoint; all state is saved
• Steer — type a new instruction while paused to redirect the agent
• Resume — continues from the checkpoint with your instruction applied
• Abort — hard stops the run immediately

Privacy

Alden is 100% local. All AI inference runs through Ollama on your own hardware. No telemetry, no data collection, no network calls except to localhost:11434.

Model Recommendations

Alden auto-detects your system RAM and recommends a model pack:

You can override the model for each phase (Enhance / Plan / Act) individually in the Settings panel.

Feedback & Issues

GitHub Issues