Can Metorial connect Langbase to AI agents?

Yes. Metorial connects AI agents to Langbase through a governed integration layer, so teams can use the provider while keeping access controlled and observable.

Does the Langbase integration work with MCP?

Metorial is MCP compatible and lets teams expose approved provider tools to MCP-capable agents and clients through a controlled access layer.

How does Metorial control access to Langbase?

Metorial applies policies across users, groups, providers, agents, and individual tools, then records the context around every agent interaction.

Can teams trace Langbase activity from agents?

Yes. Metorial records provider activity so teams can inspect tool calls, troubleshoot integrations, and give security teams the visibility they need.

Connect Langbase to AI agents

list_pipes

List Pipes

List all AI pipes (agents) in your Langbase account. Returns configuration details for each pipe including the model, status, and settings.

chunk_text

Chunk Text

Split text into smaller, manageable chunks. Useful for RAG pipelines, processing long documents, or working with specific sections of content. Uses intelligent text splitting with configurable chunk size and overlap.

append_messages

Append Messages

Append one or more messages to an existing conversation thread. Use this to add user, assistant, system, or tool messages to maintain conversation history.

create_thread

Create Thread

Create a new conversation thread in Langbase. Threads help organize and maintain conversation history across multiple interactions. You can optionally provide initial messages and a custom thread ID.

list_documents

List Documents

List all documents in a memory. Returns document names, processing status, file metadata, and chunking configuration for each document.

run_pipe

Run Pipe

Run an AI pipe to generate a response. Send messages to a configured pipe and receive the LLM completion. Supports conversation threading via threadId, variable substitution, and optional LLM provider key override.

update_thread

Update Thread

Update a conversation thread's metadata.

retry_document_embeddings

Retry Document Embeddings

Retry generating embeddings for a document that failed processing. Use this when a document's status is "failed" to re-trigger the embedding generation.

create_pipe

Create Pipe

Create a new AI pipe (agent) in Langbase. Pipes are configurable AI agents that can generate text, chat with users, and perform various AI tasks. You can configure the LLM model, system prompts, temperature, and attach memory for RAG-based responses.

web_search

Web Search

Perform a web search using the Exa search service through Langbase. Returns URLs and extracted page content for search results. Requires an Exa API key.

list_messages

List Messages

List all messages in a conversation thread in chronological order (oldest first). Returns the full conversation history including roles, content, and metadata.

generate_embeddings

Generate Embeddings

Generate vector embeddings for text chunks. Useful for semantic search, text similarity comparisons, and NLP tasks. Supports embedding models from OpenAI and Cohere.

create_memory

Create Memory

Create a new memory (RAG knowledge base) in Langbase. Memory provides vector store, file storage, and semantic similarity search. After creating a memory, upload documents to it for retrieval-augmented generation.

generate_images

Generate Images

Generate AI images using various providers including OpenAI DALL-E, Together AI Flux, Google Imagen, and more. Requires an LLM provider API key for the image generation provider.

update_pipe

Update Pipe

Update an existing AI pipe's configuration. Modify the model, prompts, temperature, and other settings of a pipe.

list_memories

List Memories

List all memory (RAG knowledge bases) in your Langbase account. Returns the name, description, and embedding model for each memory.

delete_thread

Delete Thread

Delete a conversation thread and all its messages. This action is permanent and cannot be undone.

get_thread

Get Thread

Get details of a conversation thread including its metadata.

run_agent

Run Agent

Run the Langbase runtime LLM agent directly, specifying all parameters at runtime. Supports 100+ LLMs across all major providers (OpenAI, Anthropic, Google, etc.). Unlike pipes, the agent requires specifying the model and LLM provider key per request.

delete_document

Delete Document

Delete a document from a memory. This removes the document and its embeddings permanently.

delete_memory

Delete Memory

Delete a memory (RAG knowledge base) and all its documents from Langbase. This action is permanent and cannot be undone.

web_crawl

Web Crawl

Crawl web pages to extract their content using Spider Cloud or Firecrawl through Langbase. Provide URLs to crawl and receive extracted page content. Requires a crawl service API key.

retrieve_memory

Retrieve from Memory

Perform a semantic similarity search against one or more memories. Returns the most relevant text chunks based on the query. Useful for testing RAG retrieval and debugging chunking parameters.

upload_document

Upload Document

Upload or replace a document in a Langbase memory. Langbase returns a signed upload URL and the tool uploads the content to that URL, making the document available for memory processing and retrieval.

parse_document

Parse Document

Extract text content from a document using Langbase Parser. Useful before chunking text, embedding content, or loading a document into memory.