Summary

Aligns the ScrapingBee Node.js SDK with the current documentation (llms.txt + the authoritative data/apiParameters.yml).

Added

• fastSearch() — Fast Search API
• amazonPricing() — Amazon Pricing API (wired with asin)
• gemini() — Gemini API (correctly has no search param)
• youtubeSubtitles() — YouTube Subtitles API (replaces removed youtubeTranscript())

Removed

• youtubeTranscript() (endpoint no longer supported; use youtubeSubtitles())
• youtubeTrainability() (no longer supported)

Improved

• Expanded all param type hints to be exhaustive vs the documented API contract (e.g. HTML API mode/max_cost, Google pages/date_range/geo/sort_by/price filters, Amazon Search autoselect_variant, Walmart Search start_page).
• Added the tag optional param (request label) to every endpoint's params type except Usage, including a new YouTubeMetadataParams type so youtubeMetadata() now accepts a params object.
• Fixed a pre-existing README syntax error in the HTML API POST example (missing data: key) and pointed it at httpbingo.org.
• Fixed ESLint config crash (removed the prettier/@typescript-eslint extension that was dropped in eslint-config-prettier 8).
• Bumped version to 1.9.0 and added a CHANGELOG entry.

Verification

• Unit tests: 77 passing (npm test)
• Build: clean (npm run build); dist/ in sync with src/
• Live end-to-end: all 24 integration tests pass against the real API (2 transient network/upstream failures on first pass both passed on retry — not SDK bugs)
• Programmatic cross-check: every param type is exhaustive vs data/apiParameters.yml (zero missing documented params)

Notes

• Bearer auth migration deferred to a separate PR per prior instruction.
• Pre-existing ESLint style findings (var, snake_case params, mocha globals) left as-is; only the config crash was fixed.