Integration

Agentset integration

Agentset is a managed retrieval (RAG) platform. Create namespaces (isolated knowledge bases), ingest text, URLs or files into them as ingest jobs, follow job and document processing, search a namespace with semantic or keyword retrieval and optional reranking, get download links for a document's source file or parsed chunks, and configure the hosted chat page of a namespace. The presigned file-upload routes are not exposed (the file bytes have to be PUT to the returned URL, which this catalog cannot do): ingest a public URL or text instead. Authentication is an Agentset API key sent as a Bearer token.

Connect Agentset How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in Agentset.Built from the 21 Agentset tools below, on the connection you add once.

What you can do with Agentset

Each of these is one operation on one endpoint, available both as a workflow step and as a tool an AI agent can call on its own. Nothing bundles several operations into one action, so what a run did stays readable afterwards.

Most endpoints run on the free plan. A few, mainly the ones that publish on your behalf, need a paid plan.

Namespace, 5 operations

create_namespaceCreate namespacePOSTCreates a namespace, an isolated knowledge base to ingest documents into and search. Returns its id (ns_...), passed as namespaceId to every other tool. Embedding model and vector store cannot be changed later.
delete_namespaceDelete namespaceDELETEPermanently deletes a namespace and all the data in it (ingest jobs, documents and their chunks). Cannot be undone.
get_namespaceGet namespaceGETReturns one namespace: id, name, slug, organization id, creation date and its embedding and vector store settings (third-party API keys are not returned here).
list_namespacesList namespacesGETLists every namespace (knowledge base) of the organization the API key belongs to, with its id (ns_...), name, slug and embedding and vector store settings.
update_namespaceUpdate namespacePATCHRenames a namespace or changes its slug. Send only the fields to change; the embedding model and vector store cannot be changed. Returns the namespace.

Hosting, 4 operations

delete_hostingDelete hostingDELETETurns off the hosted chat page of a namespace and deletes its hosting settings. The documents are not touched.
enable_hostingEnable hostingPOSTTurns on the hosted chat page of a namespace, a prebuilt shareable chat and search interface over its documents. Returns the hosting settings; change them with update_hosting.
get_hostingGet hostingGETReturns the hosted chat page settings of a namespace (served at app.agentset.ai/a/<slug>): title, slug, logo, system prompt, welcome message, example questions, models, topK, access protection and allowed emails.
update_hostingUpdate hostingPATCHChanges the hosted chat page settings of a namespace. Send only the fields to change; a field left out keeps its current value. Returns the settings.

Ingest job, 4 operations

create_ingest_jobCreate ingest jobPOSTQueues content for ingestion and returns the job (id job_..., status QUEUED); processing is asynchronous, so check it with get_ingest_job until COMPLETED or FAILED. Payload types: TEXT, FILE (public URL), CRAWL, YOUTUBE, BATCH.
delete_ingest_jobDelete ingest jobDELETEDeletes an ingest job together with the documents it produced and their chunks. The job moves to QUEUED_FOR_DELETE, then DELETING. Returns the job.
get_ingest_jobGet ingest jobGETReturns one ingest job: status (QUEUED, PRE_PROCESSING, PROCESSING, COMPLETED, FAILED, CANCELLED and others), error when FAILED, payload, config and timestamps. Use it to follow a job made by create_ingest_job.
list_ingest_jobsList ingest jobsGETLists the ingest jobs of a namespace, newest first, with status, payload and config. Paginated by cursor: pass pagination.nextCursor as cursor to read the next page.

Document, 3 operations

delete_documentDelete documentDELETEDeletes one document and its chunks from the namespace. Returns the document.
get_documentGet documentGETReturns one document: status, error when FAILED, source, file properties, config, chunk, token, character and page counts and processing timestamps.
list_documentsList documentsGETLists the documents of a namespace, newest first, with status, source, chunk, token and page counts. Filter by ingestJobId or statuses; paginated by cursor (pass pagination.nextCursor as cursor).

Chunks download url, 1 operation

get_chunks_download_urlGet chunks download urlPOSTReturns a temporary (presigned) URL in data.url to download the parsed chunks of a document. Only works once the document status is COMPLETED.

File download url, 1 operation

get_file_download_urlGet file download urlPOSTReturns a temporary (presigned) URL in data.url to download the source file of a document. Only for documents whose source type is MANAGED_FILE (files uploaded to Agentset).

Re ingest, 1 operation

re_ingest_jobRe ingest jobPOSTRuns an existing ingest job again with its stored payload and config (for example to pick up a changed file or website). Returns the job id; follow it with get_ingest_job.

Search, 1 operation

search_namespaceSearch namespacePOSTSearches a namespace and returns the best matching chunks in data, each with id, score (0 to 1), text and metadata. Semantic search with reranking by default; set mode=keyword for keyword search and filter to narrow by document metadata.

Warm up, 1 operation

warm_upWarm upPOSTPre-loads a namespace into the vector store's cache so the next searches answer faster. Only Turbopuffer vector stores support it. Returns data.status.

Use Agentset from Claude Code, Cursor or Codex

Connect Agentset in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its Agentset operations become tools the client can call, next to every other integration you connected.

Claude Code

claude mcp add --transport http livecontext https://livecontext.ai/mcp --header "X-API-Key: YOUR_API_KEY"

Codex (~/.codex/config.toml)

[mcp_servers.livecontext]
url = "https://livecontext.ai/mcp"

[mcp_servers.livecontext.http_headers]
"X-API-Key" = "YOUR_API_KEY"

Cursor, Claude Desktop and other JSON configs

{
  "mcpServers": {
    "livecontext": {
      "type": "http",
      "url": "https://livecontext.ai/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Set up the MCP server step by step

How to connect Agentset

Create your workspace
Sign up free. Nothing to install.
Connect Agentset once
Add your Agentset connection in Settings. Every tool from this integration reuses it, and secrets are encrypted at rest and visible only to you.
Describe the job in chat
The workflow builds itself in front of you and picks the endpoints it needs. Drag the steps yourself on the canvas whenever you would rather be explicit.