Add ChromaDB support as vector store

[ahmed-sekka]

Description

Currently only Qdrant is supported for vector storage. Add ChromaDB as an alternative, which is very popular in the RAG community.

Why ChromaDB?

• No Docker required (embedded mode)
• Popular choice for local development
• Simple API
• Good for prototyping

Tasks

• Add chromadb to optional dependencies in pyproject.toml
• Create src/core/vector/chroma.py module
• Implement ChromaVectorStore class with same interface as Qdrant
• Add --vector-db chroma option to ingest command
• Support persistent and in-memory modes
• Add integration tests
• Update documentation

Example usage

# Ingest into ChromaDB (local)
ragctl ingest chunks.jsonl --vector-db chroma --collection my-docs

# With persistence
ragctl ingest chunks.jsonl --vector-db chroma --path ./chroma_data

Files to create/modify

• src/core/vector/chroma.py (new)
• src/core/vector/__init__.py
• src/core/cli/commands/ingest.py
• pyproject.toml