The key is one thing and access to each API behind it is another: subscribe exists precisely because a fresh key meets an entitlement error on its first call to an API it was never enrolled in, and it enrols only the free BASIC tier, which never incurs charges. The descriptions also do their own routing, and it is worth following — email_search and website_contacts_scraper point at each other depending on whether you start from a query or a domain, realtime_image_search hands off to reverse_image_search or realtime_lens_data when you start from a picture instead of a word, real_time_news_search defers to realtime_news_data when you want Google News structure rather than a flat article list, and web_unblocker states outright that it is for sites no dedicated tool covers.
A front end over OpenWeb Ninja's real-time data APIs, exposing 43 tools that each take an operation name and an args object, covering Google search and AI answers, retail catalogues, real estate, jobs and pay, maps and traffic, reviews, news and video, contact discovery, and a general anti-bot fetcher.
- Google's result page as data: realtime_web_search returns organic results, knowledge panels, AI Overviews and AI Mode answers with light and bulk variants, ai_overviews and google_ai_mode fetch those answer blocks on their own, and web_search_autocomplete returns the suggestions for a partial query.
- Retail catalogues addressed by their own identifiers: realtime_amazon_data by asin, real_time_walmart_data and realtime_homedepot_data by product_id and category_id, real_time_ebay_data by product_id and seller_id, real_time_wayfair_data by sku, realtime_costco_data by product_id, and realtime_product_search/v2 across Google Shopping.
- Property listings by shape as well as address: realtime_zillow_data and real_time_redfin_data both search by location, by a coordinate box and by an arbitrary polygon, and both return property details and market trends.
- Hiring data from the angles a recruiter needs: jsearch aggregates Google for Jobs across LinkedIn, Indeed, Glassdoor and other boards, job_salary_data estimates a range for a title and location or for a named company, and realtime_glassdoor_data returns a company's reviews, salaries, interviews and open roles.
- Places and movement: local_business_data covers Google Maps search, details, reviews, photos, posts and reverse geocoding, yelp_business_data covers the Yelp side, waze returns alerts, jams and venues inside a bounding box plus directions between coordinates, driving_directions returns turn-by-turn routes, ev_charge_finder finds chargers with connector and availability detail, and local_rank_tracker measures a business's ranking across a coordinate grid built by calculate_grid_coordinates from a centre point, grid_size and radius.
- Images searched both ways: realtime_image_search filters Google Images by size, colour and type and returns full image metadata, while reverse_image_search and realtime_lens_data start from an image URL and return visual matches, exact matches, object detection and OCR text.
- Contact and reputation lookups: website_contacts_scraper pulls emails, phone numbers and social links from a domain, email_search finds addresses for a query inside an email domain and returns up to 5,000 results, social_links_search resolves a brand to its profile links, and trustpilot_company_and_reviews returns a company's Trustpilot record.
- Live text and video feeds: realtime_news_data mirrors Google News topics, sections and full-story coverage, real_time_news_search does a flat keyword search across outlets, realtime_forums_search reads community threads, real_time_video_search covers full-length video and realtime_shorts_search covers short-form, plus realtime_events_data, realtime_books_data, realtime_finance_data and play_store_apps.
- Model relays and a fallback fetcher in the same surface: chatgpt, gemini and copilot pass a prompt to those assistants, and web_unblocker retrieves the raw HTML of any URL through the anti-bot proxy network with optional JavaScript rendering.
An OpenWeb Ninja API key, supplied as OPENWEBNINJA_API_KEY. Entitlement is per API rather than per key, so a tool whose API the key has never been subscribed to answers with 401, 403 or 429 until subscribe enrols it in that API's free BASIC tier.
One command plus a key — npx -y @openwebninja/mcp-server, then supply credentials