zep-python

What it is and what it does

Zep is a client SDK for a long-term memory service designed to augment AI assistants with persistent conversation recall. It connects to a Zep backend (cloud or self-hosted) to store chat histories, automatically generate summaries, embed messages for semantic search, and retrieve contextually relevant past conversations. The SDK handles asynchronous operations to avoid blocking user interactions and provides abstractions for document vector search via Document Collections.

The package is built on three core runtime dependencies—httpx for HTTP communication, pydantic for data validation, and typing_extensions for type hints—and supports Python 3.8 through 3.12. It includes optional LangChain integration (ZepChatMessageHistory and ZepVectorStore) for use with LangChain's expression language, though that requires a separate langchain_core installation. Zep Cloud adds higher-level features like fact extraction, dialog classification, and structured data extraction from conversations.

Use it for:

• Build chatbots that recall and reference specific details from months-old conversations without storing full history in the prompt.
• Reduce token usage and latency by retrieving only semantically relevant past messages and summaries instead of full conversation logs.
• Automatically classify user intent, emotion, or segment users based on dialog patterns extracted from conversation history.
• Integrate Zep memory into LangChain or LlamaIndex applications to provide persistent context across chat sessions.
• Extract structured business data (e.g., customer preferences, order details) from unstructured chat conversations using schema-based extraction.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Zep is a long-term memory service for AI assistants that persists chat histories, generates summaries, embeds messages, and enables semantic search across past conversations to reduce hallucinations and improve context retrieval.

Yes, with conditions. The package is well-maintained for current Python versions (3.8–3.12) and has low install friction. However, dormancy for 687 days is a significant red flag: there is no recent maintenance signal, and license terms are unclear. Install only if you are committed to using Zep's backend service and can tolerate potential compatibility drift. Verify the license and test compatibility with your dependency versions before production use.

Install

zep-python on PyPI

pip install zep-python

uv add zep-python

poetry add zep-python

Installing zep-python

Before you install

Low friction install with three lightweight runtime dependencies (httpx, pydantic, typing_extensions). However, the package is dormant—no releases for 687 days as of the fact sheet date—which signals maintenance risk for bug fixes or compatibility updates with evolving Python or dependency versions.

License in practice

License treatment is unclear; no SPDX identifier or raw license text is available in the metadata. Before adopting this package in a commercial or legally sensitive context, verify the actual license terms from the project repository or documentation.

Quickstart

pip install zep-python

from zep_python import ZepClient

client = ZepClient(api_key="your-zep-project-api-key")

Requires a Zep Project API key and a running Zep service (cloud or self-hosted) to function; the SDK is a client library and cannot work standalone.

Verify before relying

• Whether the dormant maintenance status reflects active upstream development or true abandonment.
• Current compatibility of the package with modern versions of httpx, pydantic, and typing_extensions.
• Whether LangChain and LlamaIndex integrations mentioned in the description are fully functional in version 2.0.2.

Package facts

Evidence: zep_python-2.0.2-py3-none-any.whl

Tags

More Python Modules packages