A lightweight, local-first REST API simulating a core ModelVault feature: running a local LLM to respond to a prompt — fully offline.

Author: Do Minh Long ([email protected])

• POST /generate endpoint powered by a local LLM (via Ollama)
• Logs each interaction to logs/log.jsonl
• Command-line tool: minivault --prompt "..." or --status
• /status endpoint reports memory + uptime
• Docker support
• Packaged with pyproject.toml for CLI installation

• Python 3.10+
• Ollama running locally (e.g., ollama run llama3)

pip install -r requirements.txt
uvicorn minivault.main:app --reload

## Run Methods
1. Run without packaging (just scripts)
    python minivault/cli.py --prompt "Hello, who are you?"
    python minivault/cli.py --status

2. Run after packaging (via pip install -e .)
    pip install -e .
    minivault --prompt "Hello, who are you?"
    minivault --status

3. Run with Docker
    docker build -t minivault .
    docker run -p 8000:8000 minivault