LangChain Integration
Give your AI agents eyes, ears, and memory
The langchain-mixpeek package gives LangChain agents the ability to see video, hear audio, search images, and act on unstructured content. Includes a 6-tool agent toolkit (search, ingest, process, classify, cluster, alert), VectorStore with support for all file types, and bridge methods to convert between retrievers, tools, and toolkits.
Built for AI engineers and teams building agents that need to see, hear, or search unstructured content.
Get Started
Integrations
Works with any LangChain-compatible LLM and agent framework.
pip install langchain-mixpeek
npm install @mixpeek/langchainUse Cases
Brand protection agents that scan marketplaces for counterfeits
Archive search agents that find faces, objects, or moments in video
Content compliance agents that flag IP/trademark risks
Multimodal RAG pipelines across text, images, video, and audio
Features
Example
Agent with full Mixpeek capabilities
from langchain_mixpeek import MixpeekToolkit
from langgraph.prebuilt import create_react_agent
toolkit = MixpeekToolkit(
api_key="mxp_...",
namespace="brand-protection",
bucket_id="bkt_...",
collection_id="col_...",
retriever_id="ret_...",
)
agent = create_react_agent(llm, toolkit.get_tools())
result = agent.invoke({
"messages": [("user", "Scan these listings for counterfeits")]
})Agent capabilities
Tools: mixpeek_search, mixpeek_ingest, mixpeek_process,
mixpeek_classify, mixpeek_cluster, mixpeek_alert
The agent can now:
- Search video, images, audio, documents
- Upload and process new content
- Classify documents with taxonomies
- Group similar content into clusters
- Set up monitoring alerts (webhook, Slack, email)Get Started
Integrations
Details
Quick Info
Frequently Asked Questions
How does the LangChain Integration work with Mixpeek?
The integration provides a MixpeekRetriever that implements LangChain's BaseRetriever, a MixpeekTool for agent use, a MixpeekVectorStore for full CRUD, and a MixpeekToolkit that bundles 6 tools (search, ingest, process, classify, cluster, alert). When your agent calls any tool, it hits Mixpeek's API which searches across text, images, video, and audio.
What's the difference between the Python and JavaScript packages?
The Python package (langchain-mixpeek on PyPI) is the primary integration with full toolkit, VectorStore, taxonomy, cluster, alert, and plugin support. The JavaScript package (@mixpeek/langchain on npm) provides retriever and tool functionality for Node.js LangChain applications.
Can I use it with LangGraph agents?
Yes: it's the recommended approach. Use MixpeekToolkit.get_tools() with LangGraph's create_react_agent. You can scope which tools the agent gets with the actions parameter: toolkit.get_tools(actions=['search', 'ingest', 'alert']).
What file types can I ingest?
The VectorStore supports 6 blob types: text (inline strings), images (JPG, PNG, GIF, WebP), video (MP4, MOV, AVI, WebM), audio (MP3, WAV, FLAC, OGG), PDF documents, and spreadsheets (XLSX, XLS, CSV). Each has a dedicated method: add_texts(), add_images(), add_videos(), add_audio(), add_pdfs(), add_excel().
How does it compare to other LangChain vector store integrations?
Most LangChain integrations are text-only and require you to manage embeddings. Mixpeek handles multimodal embedding, indexing, and retrieval end-to-end. Plus it goes beyond search: the toolkit adds ingestion, classification, clustering, and alerting that other integrations don't offer.
Related connectors
Ready to integrate?
Get started with LangChain Integration in minutes. Check out the documentation or explore the source code on GitHub.