Search API
Search
Find relevant memories using our 5-layer hybrid search system that combines semantic understanding, keyword matching, and knowledge graph traversal.
How Search Works
Hebbrix uses a multi-layer search approach to find the most relevant memories:
Vector Search
Semantic similarity using embeddings. Understands meaning, not just keywords.
BM25 Keyword
Classic keyword matching for exact terms and names.
Graph Traversal
Finds related memories through entity relationships.
Endpoints
Code Examples
Basic Search
Python
from hebbrix import Hebbrix
client = Hebbrix()
# Simple search
results = client.search("user preferences")
for result in results:
print(f"[{result.score:.2f}] {result.content}")Search with Filters
Python
# Search within a specific collection
results = client.search(
query="project deadlines",
collection_id="col_work",
limit=20
)
# Advanced search with date range
from datetime import datetime, timedelta
results = client.search_advanced(
query="meetings",
date_range_start=datetime.now() - timedelta(days=7),
boost_recent=True
)Find Similar Memories
Python
# Get memories similar to a specific one
similar = client.search.similar(
memory_id="mem_abc123",
limit=5
)
for memory in similar:
print(f"Similar: {memory.content}")cURL Examples
POST
/v1/searchcurl -X POST "https://api.hebbrix.com/v1/search" \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"query": "What are the user's preferences?",
"limit": 10
}'Search Types
| Type | Best For | Speed |
|---|---|---|
hybrid | General queries (default) | ~50ms |
vector | Semantic/conceptual queries | ~30ms |
bm25 | Exact keyword matching | ~10ms |
graph | Entity relationships | ~100ms |