vishwa-autocomplete 1.0.0

AI-powered code autocomplete with RL-optimized suggestions

vishwa-autocomplete

Python backend for the Vishwa Autocomplete VS Code extension — AI-powered code completions with reinforcement learning.

Installation

pip install vishwa-autocomplete

Python 3.10+ is required.

Usage

This package is the backend service for the Vishwa Autocomplete VS Code extension. The extension installs and manages it automatically.

You can also run it directly:

# With a local Ollama model (default: gemma3:4b)
vishwa-autocomplete --model gemma3:4b

# With a cloud model (set your API key first)
export ANTHROPIC_API_KEY=sk-...
vishwa-autocomplete --model claude-haiku-4-5

Or as a module:

python -m vishwa_autocomplete.service --model gemma3:4b

Supported Models

Local (free, via Ollama):

• gemma3:4b — Google Gemma 3 4B (default)
• qwen2.5-coder:7b — Qwen 2.5 Coder
• deepseek-coder — DeepSeek Coder

Cloud (requires API key):

Model	Provider	Env var
claude-haiku-4-5	Anthropic	ANTHROPIC_API_KEY
claude-sonnet-4-6	Anthropic	ANTHROPIC_API_KEY
gpt-4o-mini	OpenAI	OPENAI_API_KEY

Configuration

Set your model and API keys via environment variables or a .env file:

# ~/.vishwa-autocomplete/.env
VISHWA_AUTOCOMPLETE_MODEL=claude-haiku-4-5
ANTHROPIC_API_KEY=sk-ant-...

Links

• VS Code Extension
• Source
• Issues