Agent Framework
Cline
An AI coding agent inside VS Code
Cline is an AI coding assistant that lives in your VS Code editor. It understands your entire codebase, suggests completions, writes functions, refactors code, and can make multi-file edits — all running on your local hardware. Your proprietary code never goes to GitHub Copilot's servers or Anthropic's API.
Book a consultation to set up Cline →What it is
GitHub Copilot, without sending your code to Microsoft
Cline is an open-source VS Code extension that provides AI coding assistance equivalent to GitHub Copilot — but routed to your local Ollama instance instead of Microsoft's or Anthropic's servers.
It reads your open files and understands the structure of your codebase. You can ask it to write a function, explain code, refactor a module, generate tests, or make coordinated changes across multiple files — all in plain language from the VS Code sidebar.
For companies with proprietary codebases, this is the difference between AI assistance that's viable and AI assistance that's a legal and competitive risk.
How it works
VS Code sidebar → local model → code
Cline connects to Ollama via its OpenAI-compatible API endpoint (running locally). When you ask Cline a coding question or give it a task, it reads your current file context, sends the request to Ollama, and streams the response back into the VS Code sidebar.
For multi-file edits, Cline can read files across your project, propose changes, and apply them with your approval. The model runs on your Mac; your source code never leaves your machine.
Who it's for
Developers who use VS Code and want private AI assistance
- ✓Software developers working on proprietary codebases
- ✓Freelancers under NDAs who can't send client code to cloud AI
- ✓Development teams concerned about competitive IP exposure
- ✓Anyone currently paying for GitHub Copilot who wants a local alternative
- ✓Developers who want AI assistance without monthly API subscription costs
Security note
Your proprietary code stays local. No source files are sent to Microsoft or Anthropic. The AI model runs on your Mac; the only data that leaves your machine is what you explicitly share in external integrations. This is the fundamental difference from GitHub Copilot.
Full stack
What gets installed
| Layer | Component | Purpose |
|---|---|---|
| AI Engine | Ollama (MLX backend) | Runs models on Apple Silicon |
| Agent Framework | Cline VS Code extension | AI coding agent with codebase awareness |
| Chat UI | Open WebUI | Browser chat, always available |
| Networking | Tailscale | Secure access from anywhere |
| Integration | Ollama OpenAI-compatible API | Connects Cline to local models |
| Security | Hardened config | Loopback binding, no code sent to cloud |
Security
Code never leaves your machine
Cline routes all requests to your local Ollama instance — not to any cloud API. Source files are read locally, processed locally, and edits are applied locally. We configure the Ollama endpoint to bind on loopback only. No source code is transmitted to external servers during AI inference.
Recommended models
Models that pair well
Best local coding model — designed specifically for code generation, editing, and debugging
Faster with still-excellent quality — great for completions and quick edits
Strong architectural reasoning — ideal for complex refactoring and design decisions
Ready to set up Cline?
Book a consultation. We'll install Ollama with the best coding models for your hardware and configure Cline in VS Code to use your local instance.
Book a consultation to set up Cline →