"I build things — frontend, backend, full stack, whatever you throw at me. I ask before I touch a keyboard because bad requirements make bad products. I remember everything so you don't have to repeat yourself." — Lynn
Vulpine is an intelligent agentic coding system powered by Ollama and ChromaDB. It features Lynn, a sharp and warm AI assistant designed to plan, code, and validate software projects from scratch.
Lynn isn't just a code generator. She's a builder. She:
- Clarifies: Asks smart questions before starting work.
- Plans: Breaks complex requests into actionable steps.
- Builds: Writes clean code across the stack (Python, FastAPI, React, etc.).
- Validates: Tests her own output before delivery.
- Remembers: Recalls past context across sessions using vector memory.
- Python 3.10+
- Ollama: Download Ollama
- Recommended Model:
qwen2.5-coder:7b(runollama run qwen2.5-coder:7bto pull).
Clone the repo and install dependencies:
pip install -r requirements.txtBest for auditing existing repos, fixing bugs, and generating quick docs.
python agent.pyA full-featured Streamlit interface with chat history, project context, and memory metrics.
streamlit run app.pyA floating, always-on-top PyQt5 interface that stays with you while you code.
python lynn_desktop.pyNote: Ensure lynn_api.py is running (the desktop version starts it automatically).
agent.py: Universal coding agent (CLI).app.py: Streamlit main application.lynn_api.py: FastAPI backend.lynn_desktop.py: PyQt5 floating UI.src/:agents/: Lynn's brain and core logic.core/: Modular components (Planner, Memory, reasoning, etc.).utils/: Logging and file management tools.
config/: YAML-based configuration for the agent and models.data/: Local storage for memories and knowledge base.output/: Where Lynn's creations are delivered.
Customize Lynn's personality, models, and limits in the config/ directory:
agent_config.yaml: Personality rules, identity, and capabilities.model_config.yaml: Ollama model routing and parameters.
[Insert License Here] - © 2026 Vulpine.