Get started · VS Code extension · Explore the code

Codeme connects local language models to a repository-aware development workflow. Use streaming conversations, attach code and Git context, and review proposed file changes or terminal commands before they run.

What you can explore

• Local model workflow: Ollama integration with the default qwen2.5-coder:7b model.
• Workspace context: repository browsing, ripgrep search, file previews, Git status and diffs.
• Persistent conversations: a FastAPI backend with SQLite and SQLAlchemy.
• Two interfaces: an Angular web client and a Copilot-style VS Code sidebar.
• Ask and Agent modes: focused code questions or workspace tools with visible activity.
• Explicit control: file writes and terminal commands require approval in VS Code.

Status: personal development project. Capabilities and setup are documented in the source and extension guide; this is not a hosted service or an official OpenAI product.

Getting started

Use a Python environment compatible with requirements.txt, Node.js/npm compatible with Angular 18, and a local Ollama installation. Repository search also requires rg on your PATH.

1. Prepare the backend

Run from the repository root:

python -m venv .venv

Activate with source .venv/bin/activate on Linux/macOS or .\.venv\Scripts\Activate.ps1 in PowerShell, then:

python -m pip install -r requirements.txt

Use .env.example for optional configuration. The backend defaults to SQLite in codeme.db and Ollama at http://127.0.0.1:11434.

2. Prepare the model

ollama pull qwen2.5-coder:7b

Ensure the Ollama service is running. Local inference speed depends on the available model, memory and hardware.

3. Start the API and web client

From the repository root:

python -m uvicorn main:app --host 127.0.0.1 --port 8000 --reload

In a second terminal:

cd frontend
npm install
npm start

Open http://localhost:4200. API documentation is at http://127.0.0.1:8000/docs.

Check	URL
API process	http://127.0.0.1:8000/health
Ollama/model readiness	http://127.0.0.1:8000/api/health
Application API	http://127.0.0.1:8000/api

4. Use the VS Code extension

See the extension setup, commands and configuration. Its build and packaging scripts live in vscode-extension/package.json.

Project structure

Location	Purpose
main.py	FastAPI application, lifecycle and health endpoints
codeme_agent	Model client, workspace tools, API routes and persistence
frontend	Angular web interface
vscode-extension	VS Code sidebar, editor commands and agent UI
codeme_agent/tests	Backend test suite

Development checks

python -m pytest codeme_agent/tests

Frontend build: npm run build inside frontend. Extension checks: npm run typecheck and npm test inside vscode-extension.

Run the checks in your configured local development environment.

Design priorities

Keep model execution local, ground answers in repository evidence, make tool activity visible, and keep human approval between a proposal and a consequential workspace action.