Welcome to VectorHub

VectorHub is knowledge infrastructure for AI applications. The repository shows a Robyn backend, a React application, Qdrant for vector search, Gel/EdgeDB for metadata and document workflows for parsing, chunking and retrieval.

This documentation is intentionally conservative. A capability is marked available only when it is confirmed by source code, configuration or existing documentation in the repository. Features that exist but still expose broad or unstable surfaces are marked experimental. Product direction without confirmed implementation is marked planned.

Primary investigation started from repos/VectorHub-dev/PROJECT_INDEX.md, then followed its backend, frontend, infrastructure and knowledge-gap reading paths.

Status convention

Status	Meaning
available	Confirmed by source code, configuration or runnable behavior.
experimental	Present in the repository but not stable as a public contract.
planned	Product direction or conceptual feature not yet implemented.

Verified capabilities

• Collection management through the /collections RPC endpoint and CollectionClient.
• Document upload and document processing routes in backend1/Client/server_endpoints.py.
• Semantic retrieval through POST /query, documented in VectorHub-client/README.md.
• Chunking and chunk editor workflows through /chunk and /chunk-editor.
• Metadata and file organization workflows through EdgeDB/Gel-backed clients.
• Authenticated sessions, JWT refresh and WebSocket token issuance.
• Local/private runtime using Docker Compose with API, Qdrant and Gel/EdgeDB.

What do you want to do?

• Create a collection
• Ingest documents
• Manage metadata and chunks
• Query knowledge
• Integrate with agents
• Expose knowledge through an API
• Deploy locally
• Operate private infrastructure

Main sections

• Getting Started
• API Reference
• Architecture
• Deployment
• Status and roadmap

The public landing page remains the product introduction. These docs focus on real behavior, architecture and integration.