SDKs & Libraries
Official client libraries for Python and TypeScript with type safety, auto-retries, and async support.
Python SDK
Python SDK, Python 3.8+. GitHub · PyPI
Install
pip install hebbrixFeatures
- Type Safety: Full TypeScript types and Python type hints
- Auto Retries: Exponential backoff for transient errors
- Async Support: Native async/await for non-blocking ops
Quick Start
Python
import asyncio
from hebbrix import MemoryClient
async def main():
# The Python SDK is async-first. Pair it with `async with` so the
# underlying HTTP client is closed cleanly.
async with MemoryClient(api_key="mem_sk_...") as client:
# Create a collection
collection = await client.collections.create(name="My Agent")
# Store a memory
memory = await client.memories.create(
collection_id=collection["id"],
content="User prefers dark mode",
importance=0.7,
)
# 6-layer hybrid search
results = await client.search(
query="user preferences",
collection_id=collection["id"],
limit=5,
)
asyncio.run(main())Installation Options
All Package Managers
# Using pip
pip install hebbrix
# Using poetry
poetry add hebbrix
# Using pipenv
pipenv install hebbrixAsync Usage
Python
import asyncio
from hebbrix import MemoryClient
async def main():
async with MemoryClient(api_key="mem_sk_...") as client:
# Parallel requests: the client shares a single HTTP session
# so concurrent calls reuse the connection pool.
memory, results = await asyncio.gather(
client.memories.get("mem_abc"),
client.search(query="preferences", limit=10),
)
asyncio.run(main())Error Handling
Python
import asyncio
from hebbrix import MemoryClient
from hebbrix import (
HebbrixError, AuthenticationError,
RateLimitError, ValidationError,
)
async def main():
async with MemoryClient(api_key="mem_sk_...") as client:
try:
memory = await client.memories.create(
collection_id="col_xyz",
content="Test",
)
except AuthenticationError:
print("Invalid API key")
except RateLimitError as e:
print(f"Rate limited; backend told us to slow down: {e}")
except ValidationError as e:
print(f"Request validation failed: {e}")
except HebbrixError as e:
print(f"Error: {e}")
asyncio.run(main())Configuration
Python
from hebbrix import MemoryClient
# Constructor signature
client = MemoryClient(
api_key="mem_sk_...",
base_url="https://api.hebbrix.com", # default
timeout=120.0, # seconds
)OpenAI Compatibility
Use Hebbrix as a drop-in replacement for OpenAI SDK:
Python (OpenAI SDK)
from openai import OpenAI
# Drop-in replacement for OpenAI
client = OpenAI(
api_key="mem_sk_your_hebbrix_key",
base_url="https://api.hebbrix.com/v1"
)
response = client.chat.completions.create(
model="gpt-5-nano",
messages=[{"role": "user", "content": "Hello!"}]
)
# Automatic memory injection!Available Methods
| Field | Type | Description |
|---|---|---|
| client.auth | - | register, login, create_api_key, get_me |
| client.collections | - | create, list, get, update, delete |
| client.memories | - | create, list, list_page (cursor page + metadata), iter_all (async-iterate all pages), get, update, delete |
| client.search(...) | - | 6-layer hybrid search (top-level method) |
| client.reason(...) | - | LLM-backed reasoning (top-level method) |
| client.search_resource | - | search, similar, reason (lower-level) |
| client.rl | - | train_memory_manager, train_answer_agent, get_metrics, evaluate |
| client.temporal | - | temporal-knowledge-graph access |
| client.procedural | - | procedural (skills) memory |
| client.working_memory | - | short-term context buffer |
| client.consolidation | - | memory consolidation jobs |
| client.world_model | - | world-model access |
| MemoryChat(api_key=...).send(...) | - | one-line chat integration (separate helper class) |