Menu
Mattering Automations connector

A Jina AI MCP that turns the web into LLM-ready data

Connect one API key and Jina's search foundation becomes workflow steps: read any URL as clean markdown, search the web as structured JSON, create embeddings, rerank and classify text or images, and segment long documents for free. Nothing in this connector deletes anything.

11 ready-made nodes One credential to connect Works with Claude, ChatGPT, and Matter

What the Jina AI MCP connector does

The Jina AI connector for Mattering Automations is 11 nodes across Jina's three hosts under one API key: Reader (r.jina.ai) turns any URL into LLM-ready markdown with title, description, and token usage; Search (s.jina.ai) returns web results as structured JSON; and the core API (api.jina.ai) provides embeddings, document reranking, text and image classification, classifier training, free text segmentation, and a four-node batch embedding family for large jobs (create, poll, download output, cancel).

Every node speaks MCP, the open protocol for connecting AI assistants to tools, so Claude, ChatGPT, and any MCP-capable assistant can use them, alongside Matter, the AI assistant built into the platform. An assistant can read a competitor's page as clean text, search the web for sources, and rank a pile of documents by relevance to a question, all through the same nodes your workflows use.

Because Mattering is a work platform rather than middleware, what Jina extracts lands somewhere useful by default: a page read on a schedule can update a record in a Mattering database when its content changes, search results can become research tasks with sources attached, and reranked documents can decide which item a workflow escalates to a person.

There is nothing destructive here. Every node is either a pure read or a compute call that spends API tokens and creates at most a job or a trained classifier; nothing deletes data or touches anything outside your Jina account. The one habit worth building is watching token spend: Reader and Search meter by tokens processed, and every response carries a usage figure so workflows can track exactly what a run cost.

One connection, three surfaces

  • Visual automations. Drag Jina AI nodes onto the canvas and wire them to triggers, conditions, and other apps.
  • AI agents, yours or ours. These nodes speak MCP, the open protocol, so Claude, ChatGPT, and any MCP-capable assistant can use them, and Matter, the agent built into the platform, uses them natively.
  • The whole platform. Combine with native tasks, projects, and other connectors, no middleware in between.

Every Jina AI node, ready to drop on the canvas

11 nodes, generated from the connector itself so this list always matches what ships. Expand a category to browse.

Read and query nodes 5

Jina: Download Batch Output

Download the finished vectors of a completed batch embedding job.

Jina: Get Batch Job

Poll a batch embedding job's status and progress until it completes or fails.

Jina: Read URL

Fetch any URL through Jina Reader (r.jina.ai) and return its title, description and main content as LLM-ready markdown, with token usage. Steering options control return format (markdown, html, text, screenshot, pageshot), CSS target selectors, token budget and image handling. Screenshots return as hosted URLs.

Jina: Search Web

Search the web through Jina Search (s.jina.ai) and return results as structured JSON: title, URL, description and content per result, with site restriction and result count options.

Jina: Segment Text

Split long text into chunks sized for embedding models or LLM context windows. Free: costs zero tokens, yet still requires a valid API key, which makes it the connector's auth probe.

Action nodes 6

Jina: Cancel Batch Job

Cancel a running batch embedding job that is no longer needed.

Jina: Classify Text or Image

Classify text or image inputs against label names zero-shot, or against a trained classifier by id.

Jina: Create Batch Embedding Job

Submit a large corpus as one asynchronous batch embedding job; returns a job id immediately. Poll with Jina: Get Batch Job and retrieve vectors with Jina: Download Batch Output.

Jina: Create Embeddings

Turn text or images into vectors synchronously. For thousands of inputs, use the batch embedding job family instead.

Jina: Rerank Documents

Order candidate documents by relevance to a query, with scores. The quality step between a wide vector-search shortlist and the documents an LLM or person actually sees.

Jina: Train Classifier

Train a classifier on labeled examples and return a classifier id that subsequent classify calls reference. Jina does not document listing or deleting classifiers, so store the returned id durably.

Connecting Jina AI takes one credential

Jina issues API keys from jina.ai (the dashboard shows the key and its remaining token balance). One key works across all three hosts the connector uses: api.jina.ai, r.jina.ai, and s.jina.ai. Mattering sends it as a bearer credential on every request.

1

Get an API key from jina.ai and note its token balance.

2

In Mattering Automations, open Connectors, click Add Connector, choose Jina AI, and create a credential with that key.

3

Click Test Connection. Mattering calls Jina's free Segmenter endpoint, which costs zero tokens yet still rejects invalid keys, so the test is a true auth check that spends nothing.

4

All 11 Jina nodes appear in the canvas palette. Jina: Read URL on any public page is a good first real run: the response shows the markdown content and the exact token usage.

Common uses for the Jina AI MCP

Any URL as clean, LLM-ready markdown

Jina: Read URL fetches a page and returns its title, description, and main content as markdown, with token usage attached. Steering options map to Jina's own controls: return format, target selectors, token budget, and image handling. Screenshots come back as hosted URLs rather than raw bytes, so workflow payloads stay light.

Web search that returns data, not HTML

Jina: Search Web queries the web and returns results as structured JSON: title, URL, description, and content per result. Site restriction and result count are plain config fields, which makes it a natural first step for research workflows that fan out into Reader calls on the best hits.

Semantic search over your own content

Jina: Create Embeddings turns text or images into vectors, and Jina: Rerank Documents orders candidate documents by relevance to a query. Together they cover the retrieval half of RAG: embed once, then rerank the shortlist at question time and hand only the top documents to a model.

Classification with or without training

Jina: Classify Text or Image labels inputs against label names zero-shot, no training required. When the labels are stable and accuracy matters, Jina: Train Classifier fits a classifier on labeled examples and returns a classifier id that subsequent classify calls reference.

Free segmentation for long documents

Jina: Segment Text splits long text into chunks sized for embedding models or LLM context windows, and it costs zero tokens. Run it ahead of embedding jobs so chunk boundaries are deliberate rather than arbitrary truncation.

Batch embeddings for large corpora

For thousands of texts, Jina: Create Batch Embedding Job submits the whole corpus as one asynchronous job. Jina: Get Batch Job polls status, Jina: Download Batch Output retrieves the finished vectors, and Jina: Cancel Batch Job stops a job that is no longer wanted. A scheduled poll loop on the canvas handles the waiting.

Example: A research question becomes ranked, readable sources in a task

Built on the Automations canvas in a few minutes, or described in plain language to Matter and built for you.

1

Trigger: a task created in Mattering Productivity carries a research question.

2

Jina: Search Web queries the question and returns structured results with URLs.

3

Jina: Read URL runs on the top hits, turning each page into clean markdown with token usage tracked per page.

4

Jina: Rerank Documents orders the page contents by relevance to the original question.

5

The top documents and their sources land back on the task as a comment, so a person reviews actual content, not a pile of links.

Jina AI MCP questions, answered

What Jina AI operations are supported?

11 nodes: Reader (URL to markdown), web search as JSON, embeddings, document reranking, text and image classification, classifier training, free text segmentation, and batch embedding jobs (create, poll status, download output, cancel). Deliberately not shipped: DeepSearch, which is streaming-first and runs for minutes, so it belongs in a longer-lived worker rather than a canvas node; EU data residency hosts; and classifier listing or deletion, which Jina does not document.

Does this work with Claude, ChatGPT, or other AI assistants?

Yes. Mattering exposes these nodes over MCP, an open protocol for connecting AI assistants to tools, so any MCP-capable assistant, including Claude and ChatGPT, can connect to your workspace and use them. Matter, the assistant built into the platform, uses the same nodes natively.

Can anything in this connector delete or damage data?

No. Every node is either a pure read or a compute call: the strongest side effects in the whole connector are creating a batch job, training a classifier, and canceling a batch job you created. Nothing deletes data, and nothing touches systems outside your Jina account. The practical thing to watch is token spend, and every response includes a usage figure for exactly that.

What does it cost to run these nodes?

Jina meters by tokens processed against your API key's balance, and the connector never invents prices: check jina.ai for current rates. Two useful facts for budgeting: Jina: Segment Text is free (zero tokens, per Jina's own docs), and Reader and Search responses carry a usage.tokens figure, so a workflow can log exactly what each run consumed. Reader also accepts a token budget option to cap how much of a large page is processed.

How does the connector handle Jina's three different hosts?

Transparently. Reader lives at r.jina.ai, Search at s.jina.ai, and everything else at api.jina.ai; each node already knows its full URL and the same API key authenticates all three. Nothing to configure.

Can Reader take screenshots of pages?

Yes: set the return format to screenshot (visible area) or pageshot (full page). Because the connector always requests Jina's JSON envelope, the screenshot arrives as a hosted URL rather than raw image bytes, which keeps node outputs small and gives downstream steps a link they can store or send.

How do the batch embedding nodes fit together?

As an asynchronous create-poll-download chain. Jina: Create Batch Embedding Job submits the corpus and returns a job id immediately. Jina: Get Batch Job reports progress until the job completes, typically inside a scheduled polling loop. Jina: Download Batch Output then retrieves the vectors, and Jina: Cancel Batch Job abandons a job early if the workflow no longer needs it. For a handful of texts, plain Jina: Create Embeddings is simpler and synchronous.

Why is the connection test free?

Because Jina's Segmenter is the rare endpoint that costs zero tokens yet still validates authentication: an invalid key gets a 401 even though valid calls are unmetered. The connector uses it as the connection test, so testing a credential never spends from your balance.

11 nodes waiting on the canvas

Put Jina AI to work inside your workspace

Connect once, then use every node in visual automations and AI agent workflows across the Mattering platform.

All product names, logos, and brands are property of their respective owners and are used for identification purposes only. Use of these names, logos, and brands does not imply endorsement or affiliation. Jina AI is a product of its respective owner; Mattering is an independent platform.

Start Free Explore our Products