Extract high-value keyword candidates from any HTML page using Readability parsing, POS tagging, and frequency analysis.
Get StartedMozilla Readability strips boilerplate (nav, ads, sidebars) to isolate the main article content for accurate analysis.
compromise identifies nouns, proper nouns, and noun phrases. Multi-word phrases like "search engine optimization" are detected automatically.
Candidates scored against a 50,000-word English frequency corpus. Uncommon terms rank higher as stronger SEO targets.
Results cached by domain + content MD5. Same content returns instantly from cache, regardless of when it was last fetched.
Companion plugin auto-extracts keywords on publish and assigns them as taxonomy terms for internal linking and SEO.
Simple POST endpoint accepts raw HTML and returns structured keyword candidates with stats. Integrate with any CMS or pipeline.
POST raw HTML to the extraction endpoint, optionally with the source URL for domain-based caching.
Readability isolates article text. compromise tokenizes and tags parts of speech. Stopwords and short terms are filtered out.
Terms scored by inverse frequency (rare = better SEO target). Duplicates merged by lowercase root form.
Structured JSON response with each candidate's term, rank, phrase status, and extraction stats.
Single-endpoint extraction — send HTML, receive keywords.
POST /api/extract
Content-Type: application/json
{
"html": "<html>...</html>",
"url": "https://dealer.example.com/specials"
}
Response:
{
"title": "Summer Service Specials",
"candidates": [
{ "term": "oil change", "rank": 12400, "isPhrase": true },
{ "term": "brake inspection", "rank": 28000, "isPhrase": true }
],
"stats": { "raw": 145, "filtered": 38, "candidates": 12 },
"cached": false
}