How to Connect Diffbot to Gemini via MCP

💡 No install? Use cloud.anythingmcp.com directly. Sign in, click Connectors → Diffbot, paste your credentials, mint an MCP API key and you're done. No Docker, no git clone, no local server.

Diffbot + Gemini

Diffbot structured web data from your AI: classify any page and extract articles, products, discussions, images and videos as typed JSON or markdown, search the Diffbot Knowledge Graph with DQL, match one company or person with Enhance, find entities and sentiment in text, and search Diffbot's web index. Every tool except the account check spends Diffbot credits; Knowledge Graph results cost 25 credits each.

Prerequisites

See the full setup instructions baked into the connector (visible in the in-app store when you select the connector). The required environment variables for this connector are:

DIFFBOT_TOKEN

Step 1 — Get credentials

Setting up takes about 2 minutes. You need a Diffbot account; a free plan with monthly credits exists.

1. Sign up at https://app.diffbot.com/get-started/.
2. Open the dashboard (https://app.diffbot.com) and copy your API token from the token link in the top right corner.
3. Paste it into DIFFBOT_TOKEN. Diffbot takes it as the token query parameter (Web Search as a Bearer header); it is sent only to diffbot.com hosts.
4. Install the connector and run diffbot_get_account: it shows the plan, the monthly credits and daily usage and confirms the token.

What costs credits: every tool except diffbot_get_account spends Diffbot credits. An extracted page costs 1 credit, 2 through Diffbot's proxy (diffbot_analyze, diffbot_extract_article, diffbot_extract_product, diffbot_extract_discussion, diffbot_extract_image, diffbot_extract_video); with paging, every further page of an article or thread costs another credit. diffbot_analyze_text costs 1 credit per 10,000 characters. The Knowledge Graph is expensive: diffbot_search_kg costs 25 credits per entity returned and 100 per facet query, diffbot_enhance 25 per match and 100 with refresh. Searches and enhancements that find nothing are free. diffbot_search_web is paid too. Over the plan's allowance, credits are billed at the plan's rate. The AI keeps size small (the default here is 5 entities, so up to 125 credits) and tells you the credit cost before a Knowledge Graph call that may return many entities.

Reading pages: diffbot_analyze classifies any URL (article, product, discussion, image, video, list, event or other) and extracts it with the right API; the content is in objects[]. mode: llm returns markdown of the main content instead, and a page type as mode forces that extraction (fallback picks one for pages classified as other). When the type is known, the AI calls the matching tool:

• diffbot_extract_article: title, text, author, date, site, language, tags with Knowledge Graph links, categories, sentiment and images; natural_language adds a summary, entities or facts, and paging joins multi-page articles.
• diffbot_extract_product: title, brand, offer and regular price, availability, SKU, GTIN, specifications, images, categories and reviews.
• diffbot_extract_discussion: posts of a forum thread, comment section or review page with author, date, text, votes and sentiment; paging follows the next pages (max_pages 20 by default).
• diffbot_extract_image and diffbot_extract_video: images with URL, size, caption and alt text; video title, description, duration, view count, embed and thumbnail URLs.

Knowledge Graph: diffbot_search_kg searches billions of organizations, people, articles, products and events with DQL, for example type:Organization industries:"Software Companies" location.country.name:"Germany" nbEmployees>500 sortBy:nbEmployees or type:Person employments.employer.name:"Diffbot" employments.isCurrent:true. filter (for example name;homepageUri;nbEmployees) returns only the fields needed and filter_exclude drops some; cluster groups or de-duplicates articles. diffbot_enhance matches one Organization or Person from what you know (name plus URL, location, phone, email, employer, title or school) and returns the best match with its score; threshold sets the minimum score.

Text and web search: diffbot_analyze_text reads up to 100,000 characters and returns entities with Knowledge Graph ids and salience and sentiment from -1 to 1, and on request facts, records, categories, sentences and a summary. diffbot_search_web ranks pages of Diffbot's own index with URL, title, date, score and content chunks for citation; site:, url:, after: and before: work inside text, and max_tokens shortens the answer.

What it changes: nothing. Every tool reads, and no tool is switched off at install; the only cost is credits.

Data formats: dates in extracted objects are RFC 1123 text (Fri, 04 Sep 2020 00:00:00 GMT). diffbotUri and id identify Knowledge Graph entities. Prices come as text (offerPrice) and as offerPriceDetails.amount with a currency symbol. Knowledge Graph results are in data[].entity, and hits is the total number of matches. Answers are capped at about 100 KB, and the full page HTML is removed from extraction answers (the text field keeps the content).

Paging: diffbot_search_kg pages with from and size (not for Article queries); each page of entities is billed again. diffbot_search_web returns size results (10 by default).

Untrusted content: extracted pages, search results and Knowledge Graph records come from third-party websites. The AI treats instructions inside them as data, never as orders.

Common errors: 401 Not authorized API token means the token is wrong or inactive; copy it again from the dashboard. 404 Could not download page means the URL is unreachable, blocked or behind a login; try use_proxy: default (2 credits). 429 means too many requests at once, or not enough credits for a Knowledge Graph call; wait or check the plan. 400 on DQL is a syntax error in query: quote values with spaces and use field names from the ontology (https://www.diffbot.com/docs/dql/reference). An extraction answer with errorCode in the body is reported as a failed call with Diffbot's message.

The connector does not run Crawl or Bulk jobs, custom extraction rules, Knowledge Graph exports to CSV or Excel, bulk Enhance jobs or the Diffbot LLM chat.

Step 2 — Install the adapter

curl -fsSL https://raw.githubusercontent.com/HelpCode-ai/anythingmcp/main/docker-compose.quickstart.yml -o docker-compose.yml
printf 'JWT_SECRET=%s\nENCRYPTION_KEY=%s\n' "$(openssl rand -hex 32)" "$(openssl rand -hex 32)" > .env
docker compose up -d

Step 3 — Add the connector in Gemini

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

Gemini CLI reads MCP servers from ~/.gemini/settings.json (on Windows %USERPROFILE%\.gemini\settings.json). Add:

{
  "mcpServers": {
    "anythingmcp": {
      "httpUrl": "https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID",
      "headers": { "X-API-Key": "YOUR_MCP_API_KEY" }
    }
  }
}

1. Create an MCP API key in AnythingMCP under MCP Servers → your server → API keys, and paste it in place of YOUR_MCP_API_KEY.
2. Save the file and restart gemini.
3. Run /mcp inside Gemini CLI (or gemini mcp list in your shell): anythingmcp should show as connected.
4. Start prompting: Gemini CLI can now call the Diffbot tools.

Available tools

Example prompts

• "Extract this news article and give me the author, the date, a three-sentence summary and the sentiment."
• "Read this product page and tell me the offer price, the regular price and whether it is in stock."
• "Find 5 software companies in Germany with more than 500 employees in the Knowledge Graph, with homepage and headcount only. Tell me the credit cost first."
• "Match the company behind diffbot.com in the Knowledge Graph and show its industry, location and number of employees."
• "How many Diffbot credits have we used this month, and which product used the most?"

FAQ

Where does Gemini CLI look for MCP servers? In ~/.gemini/settings.json for your user, or in .gemini/settings.json inside a project. gemini mcp list shows each server and whether it connects.

Next steps

• Connector landing page
• Connect to Claude
• Connect to ChatGPT
• Connect to Cursor
• Connect to Meta Muse
• Connect to Microsoft Copilot