The src/ tree houses the Von runtime, tooling, and research assets. Use this guide to understand the responsibilities of each area and where to extend functionality.

Full server stack for the Vontology platform.

• server/ Flask app, route handlers, and helpers exposed to the UI and MCP clients.
• services/ Core domain logic for concepts, annotations, relation elicitation, chat history, similarity, and text values (with focused tests in services/tests).
• db/ Mongo connection management plus repositories for concept, relation, and text-value persistence.
• integrations/ Bridges to internal/external MCP tools (arXiv proxy, orchestration gateway) and JIRA automation.
• languagemodels/ Shared LLM interface plus Ollama/OpenAI clients and utilities.
• mcp_server/ HTTP and stdio MCP servers kept in sync with the internal gateway catalogue.
• models/, prompt/, security/, utilities/, utils/, and vontology/ provide schema definitions, prompt templates, access control, maintenance scripts, general helpers, and Vontology-specific utilities respectively.

Currently focuses on the web/von_interface Flask UI, with templates/ for Jinja views and static/ assets for the embedded terminal-inspired experience.

Curated, versioned Vontology source material (knowledge/vontology) that seeds and documents concept hierarchies for research workflows.

Cloud-facing memory back ends (memories/cloud), currently geared towards Google Drive synchronisation hooks.

Standalone operational scripts for maintainers, including local database initialisation, Ollama control helpers, remote test harnesses, and cleanup utilities.

Composable workflow definitions (e.g., onboarding_workflow.py and von/main.py) that orchestrate services, memories, and personas for specific research tasks.

• GEMINI.md captures recent tool-usage learnings for Gemini-related automation.
• __init__.py marks the package root so modules under src/ can be imported using absolute paths.