PrismCrawl

Changelog

  • Improved Google Search compatibility with existing encoded locations (`uule`), so more saved location values can be reused without regenerating them.
  • Fixed an issue that caused Walmart Reviews requests for some products to fail even when reviews were available.
  • Fixed MCP connection failures in Claude and Claude Code when signing in or using an API key. Restart the connection if an earlier attempt failed.
  • Added Google APIs for flights, hotels, travel exploration, finance, trends, patents, Scholar, images, videos, events, jobs, and additional Maps data, with source-specific controls and pagination. Flight searches support adults, children, infants in seats, and infants on laps.
  • Added search and product data across Bing, Baidu, DuckDuckGo, eBay, Walmart, Yahoo, Naver, and other providers, plus YouTube search, channels and video details, Zillow listings, and OpenTable reviews.
  • The API documentation, dashboard request builder, request history, and MCP tools cover the new services, including their supported parameters and request replay.
  • Invalid place/review identifiers, unsupported travel currencies, invalid Trends comparison terms or date ranges, and impossible flight dates or passenger combinations return HTTP 400 before processing.
  • Google sponsored results preserve an optional `tracking_url` and expose the advertiser destination in `url` when supplied by the source.
  • Google Play Reviews and Product requests with malformed IDs now return a clear HTTP 400 validation error before processing. Use the ID from Google Play search results or the product URL.
  • Added Connected apps to view and disconnect individual MCP OAuth connections. Connection approval now explains continued access and credit usage, with a return path after API-key setup.
  • Added Google Search parameters for raw encoded locations (`uule`), personalization preference (`pws`), and already URL-encoded queries (`query_encoded`).
  • Released official Python and Node.js SDKs: pip install prismcrawl and npm install prismcrawl. Both cover every endpoint, with types, automatic retries, and pagination helpers.
  • Published the PrismCrawl MCP server to the official MCP Registry as io.github.prismcrawl/prismcrawl and listed it on Glama.
  • Added setup instructions for Claude Code, Cursor, and VS Code to the MCP page.
  • Improved AI Overview extraction rate at no additional cost.
  • New demo accounts now receive 100 free credits, up from 25, with no credit card required. Existing account balances are unchanged.
  • Added optional coordinate location targeting to Bing Search. Latitude and longitude are forwarded as Bing's native client-location signal, echoed in normalized search parameters, retained in request history, and bound to continuation tokens.
  • Fixed an intermittent Google Search query mismatch that could cause some multi-word searches to return results for only their first word.
  • Google featured snippets now include their source page title, snippet text, and source URL in normalized SERP features.
  • Added delivery_country, postal_code, and currency controls to Amazon Search. Marketplace, interface language, delivery destination, and display currency can now be selected independently, with marketplace-derived defaults.
  • Added normalized Google News, Books, and Local search responses with article source and recency metadata, book results and descriptions, and structured local-result cards.
  • Expanded Google Maps results with service-area businesses that do not publish coordinates, Plus Codes, highlighted review excerpts, popular-times data, additional named hours, and broader descriptions, category identifiers, and imagery.
  • Added Amazon Search with normalized product, price, rating, promotion, delivery, filter, category, related-search, and pagination data. Requests can target supported Amazon marketplaces, languages, devices, pages, categories, sorts, refinements, and merchants, or return encoded source HTML.
  • Added the optional google_domain parameter to Google Search. Choose a search-capable Google domain, such as google.co.uk; omitted values continue to use google.com, independently of country, language, and location controls.
  • Expanded Google Search controls with coordinate radius targeting, tablet requests, native tbm and udm search tabs, and broader tbs advanced filters for sorting and search-tab-specific filtering.
  • Expanded Google Maps results with language-independent category IDs, international phone numbers, listing claim state, highlighted details, ordering and reservation links, plus hotel pricing, booking, class, check-in, check-out, and amenity details.
  • Added Google Maps search with coordinate and zoom targeting, 20-place offset pagination, and normalized place IDs, rankings, addresses, coordinates, categories, ratings, review counts, contact details, hours, imagery, and business attributes.
  • Added source_name to normalized Google and Bing results, using the site/source label displayed by the search engine and returning null when no label is available.
  • Stabilized normalized JSON response shapes: unavailable scalar and object values now return null, while evaluated empty collections return [], so result and SERP feature fields no longer appear conditionally.
  • Added safe and color_scheme controls to Google and Bing, Bing's native sp hint, and Google's native nfpr, filter, lr, and cr parameters, including normalized response metadata, auditing, and Bing continuation-token validation.
  • Added the optional zero_trace parameter to Google and Bing search requests. Zero-trace requests retain audit and billing metadata but do not archive source HTML or parsed JSON; request history exposes only the response status and whether credits were charged.
  • Added optional tbs time-range filtering to the Google and Bing search endpoints, including relative and custom ranges plus filtered Bing pagination. Requests that omit tbs continue without a time filter.
  • Improved SERP feature extraction across non-English result pages for more consistent localized data.
  • Request metadata is now retained for 90 days, with paired HTML and JSON artifacts available for 30 days.
  • Added authenticated dashboard and API access for searching past requests, inspecting request details, and downloading retained artifacts.
  • Launched this changelog so new PrismCrawl features and improvements are easy to follow.