ChromaDB Retriever Integration

[github-actions]

Description

Add a ChromaRetriever backed by ChromaDB, a popular open-source vector database with persistent storage and metadata filtering.

Motivation

ChromaDB provides on-disk persistence, rich metadata filtering, and a simple Python API — ideal for prototype-to-production workflows where data must survive process restarts.

Acceptance criteria

• Class ChromaRetriever in ragframework/retriever/chroma.py
• Subclasses Retriever from ragframework/base.py
• Uses chromadb (already listed in [chromadb] optional extra in pyproject.toml)
• Supports both ephemeral (client = chromadb.Client()) and persistent (chromadb.PersistentClient(path=...)) modes via constructor argument
• Import guarded; raises RetrieverError on failure
• Unit tests using ephemeral mode (no disk I/O)
• Exported from ragframework/retriever/__init__.py
• CHANGELOG.md updated under [Unreleased]

Files to touch

• ragframework/retriever/chroma.py — new file
• ragframework/retriever/__init__.py — export it
• tests/test_retriever/test_chroma.py — new test file

Resources

• ChromaDB docs
• InMemoryRetriever in ragframework/retriever/in_memory.py as reference

[VanshikaLud04]

I'd like to take this on! I'm familiar with ChromaDB and have worked with LangChain before .