Extract SEO keywords from any text content

Keyword Extractor pulls up to 50 relevant SEO keywords from a piece of text and returns them as exactly one comma-separated, all-lowercase line, ordered by relevance, with no numbering, bullets, explanations, or trailing period. It works in five steps: analyze the text for its main subject, key topics, domain terminology, entities, and concepts while ignoring filler words; generate up to 50 keywords as single words or 2-4 word phrases; rank them by SEO importance (main topic first, then high-value domain terminology, then technologies/tools/entities mentioned, then common search queries, then supporting context); normalize by removing duplicates and near-duplicates, keeping only the more common search phrase when two keywords represent the same concept; and validate the final list against the same rules before returning it, regenerating the whole list if any rule fails rather than patching individual entries.

Use it for any request to extract or generate SEO-friendly keywords or tags from text - for search, metadata, or tagging - regardless of how the request is phrased, as long as the goal is keyword or tag generation. It should not trigger for summary or paraphrasing requests, or for text analysis that does not itself ask for keyword generation. Keywords are meant to be phrases a real user would type into a search engine - the rules explicitly reject weak keywords like "things," "various topics," "general concepts," or "important ideas," and prefer noun phrases and domain terminology over verbs or adjectives:

machine learning, neural networks, deep learning models, ai algorithms, data science tools

Input is any text or paragraph. Output is a single line of up to 50 comma-separated, lowercase keywords with no duplicates or near-duplicates, each no longer than four words, ranked most important first. If the input text is very short, the skill still infers likely topics and generates keywords rather than returning nothing or refusing, but it never exceeds 50 keywords regardless of how much text it is given.

None - it is a pure text-in, text-out transformation with no external service calls, API keys, or file access involved.

Anyone producing SEO metadata, content tags, or search-facing keyword lists from arbitrary text - content teams doing on-page SEO, tooling that auto-tags articles or products, or an agent pipeline that needs a consistent, machine-parseable keyword line rather than a prose summary of what a document is about.