Skip to main content

Knowledge Store API

ChromaDB-based vector storage for RAG.

KnowledgeStore​

from owl.knowledge import get_knowledge_store

store = get_knowledge_store()

Methods​

add_document()​

Add a document to the knowledge base.

result = store.add_document(
    file_path="/path/to/doc.md",
    project_path="/path/to/project"  # Optional
)
# {"filename": "doc.md", "chunks": 12}

Semantic search.

results = store.search(
    query="authentication flow",
    project_path="/path/to/project",  # Optional filter
    limit=3
)
# [{"text": "...", "filename": "...", "relevance": 0.85}, ...]

remove_document()​

Remove a document.

removed = store.remove_document("/path/to/doc.md")
# True/False

get_stats()​

Get knowledge base statistics.

stats = store.get_stats()
# {"total_chunks": 45, "sources": [...]}

list_sources()​

List all document sources.

sources = store.list_sources()
# [{"filename": "doc.md", "chunks": 12, "project": "..."}, ...]

Configuration​

Embedding Model​

Default: nomic-embed-text

Ensure it's available:

ollama pull nomic-embed-text

Storage Location​

~/.owl/knowledge/chroma/

Example Usage​

from owl.knowledge import get_knowledge_store

store = get_knowledge_store()

# Add documentation
store.add_document("README.md")
store.add_document("docs/api.md")

# Search
results = store.search("how to authenticate")
for r in results:
    print(f"{r['filename']}: {r['text'][:100]}...")

# Remove
store.remove_document("docs/api.md")