Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Micro-Tool Niche Finder: Score Saved Keyword and SERP Data

    1

    A short list of keywords for a small free tool page, made from numbers in keyword and SERP files you saved. It calls no API and asks for no key.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Micro-Tool Niche Finder: Score Saved Keyword and SERP Data

    Micro-Tool Niche Finder: Score Saved Keyword and SERP Data

    Example session with this skill installed

    Here is my keyword ideas export and the saved results page for "free bmi calculator". Which of these could I build a free calculator page for?

    • Read your context and instructions
    • Compiled the micro-tool niche finder

    Summary

    The skill's own script ran on the example files shipped in the ZIP (references/example): 3 candidates, 2 skips, 1 warn, 0 unknown.

    $ python3 scripts/niche_finder.py --keywords references/example/keyword-ideas.json --serp references/example/serp-free-bmi-calculator.json --out ./niche-out
    files read=2 unknown=0
    output: NICHE-SHORTLIST.md and niche-shortlist.json in the --out folder
    NICHE candidate=3 skip=2 warn=1 unknown=0          (exit 0)
    

    From NICHE-SHORTLIST.md:

    candidate (highest score first)

    1. tile spacing calculator: candidate, score 97.78 (search volume 880, keyword difficulty 8); SERP: not given
    2. free bmi calculator: candidate, score 87.50 (search volume 1400, keyword difficulty 15); SERP: warn, forum or user-generated pages in the top 10: 2 (reddit.com, quora.com; organic results counted: 10, distinct domains: 9)
    3. paint coverage calculator: candidate, score 26.82 (search volume 590, keyword difficulty 21); SERP: not given

    skip

    • loan payment calculator: skip (search volume 2400, keyword difficulty above 30); SERP: not given
    • unit converter for bakers: skip (search volume below 500, keyword difficulty 5); SERP: not given

    A second run on the failing example shows that nothing passes when the numbers do not:

    $ python3 scripts/niche_finder.py --keywords references/example-failing/keywords.csv --out ./niche-out-2
    NICHE candidate=0 skip=3 warn=0 unknown=0          (exit 1)
    

    Next steps

    • Open the results page yourself for any keyword marked warn.
    • Pick a candidate and build the tool page; the skill does not build it or forecast its traffic.
    From NICHE-SHORTLIST.md

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Use when you have exported keyword ideas or a simple keyword CSV and want to know which keywords to build a calculator or converter page for, when you want to see which of your saved results pages that can be read have forum or user-generated pages in the top 10, or when an agent should suggest tool ideas but first needs a filtered list. A standard-library script reads only the files you give it (a keyword ideas JSON export in the field layout written out in the rules file, or a CSV with a fixed three-column header, plus optional SERP JSON files) and calls no API, asks for no key and fetches nothing. Each keyword gets a demand and a competition outcome from two thresholds you can change (a search volume of at least 500 and a keyword difficulty of at most 30 by default), and a keyword is a candidate only when both numbers can be read and pass; a file or value that cannot be read gives unknown, never candidate. Candidates are sorted by a fixed score, search volume / (keyword difficulty + 1), and only numbers that can be read get one; a keyword that is not text is unknown. When a saved results page that can be read shows 2 or more forum or user-generated pages among the organic top 10, the keyword gets warn as a hint to look yourself; warn changes neither the status nor the score. Keywords that pass a text rule are shown as they are; a forum or user-generated domain from a file that can be read is shown only as the matching name from the skill's list, and titles, descriptions, URLs, domain texts and other fields are never shown. Do not use when you want keyword data collected, a content plan, an income estimate or the tool page itself.

    Limits

    A candidate row means that two numbers read from your file passed two thresholds, and a value that cannot be read gives unknown instead; it is not a forecast of visits or income and says nothing about whether a page can reach the first results page. The skill cannot tell whether the numbers are recent, cover every keyword, or which country and language they are for, and it does not use cost per click or search intent. The thresholds, the forum and user-generated domain list, the warn rule and the score are heuristics applied only to values that can be read; what the files contain comes only from two documentation pages: field names, and one quoted sentence on keyword difficulty. A keyword that passes its text rule is shown exactly as it is, so a private word in a keyword appears in the output. It does not collect data, write content plans or estimate income.

    Requires

    Python 3.9+ (standard library only; tested with 3.9). You give the saved files and an output folder outside the folders that hold them; the script reads only those files and its own patterns file (an input file that is itself a link is not followed; a linked parent folder is followed), creates the output folder (and missing parent folders), writes only NICHE-SHORTLIST.md and niche-shortlist.json there, starts no process and makes no network calls.

    Example prompts

    1. "Here is my keyword ideas export. Which of these could I build a free calculator page for?"
    2. "Shortlist the keywords in this CSV with at least 1,000 searches and a keyword difficulty of 20 or less."
    3. "I saved the results pages for my top keywords too. Which of them show user-generated pages in the top 10?"

    Independent

    Independent tool. Not affiliated with or endorsed by DataForSEO, Google, Reddit, Quora, Stack Exchange or Medium, nor by Stack Overflow, Anthropic, OpenAI, Anysphere (Cursor) or the Python Software Foundation. Product names are used only to describe compatibility, the file format the skill reads and the kinds of sites it counts.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 1 day ago

    Requires

    Python 3.9+

    Permissions

    Terminal / Shell
    Read Files
    Write Files

    Runs one Python 3 standard-library script locally. It reads only the keyword and SERP files you give it, writes NICHE-SHORTLIST.md and niche-shortlist.json to the --out folder (outside the input folders), makes no network calls, needs no API keys or env vars and starts no process.

    Listed1 day ago

    What's inside

    Frequently Asked Questions

    No. candidate means two numbers read from your file passed two thresholds (search volume at least 500 and keyword difficulty at most 30 by default). It says nothing about visits, income or whether a page can reach the first results page.

    A file, a CSV row or a value could not be read, or the keyword is not text. It never turns into ok or candidate, and a keywords file that cannot be read lists no keyword at all.

    Only in the SERP part: a saved results page that can be read shows 2 or more forum or user-generated pages among the organic top 10. It is a hint to open the results page yourself and changes neither the status nor the score.

    No. It reads only the saved files you give it, makes no network call and starts no process. Never paste an API key, a password or a login into the chat; export newer data yourself and give the saved file.

    Only keywords that pass the keyword rule, exactly as they are, so a private word in a keyword appears in the output. Titles, descriptions, URLs and other fields are never shown, and a forum domain is shown only as the matching name from the skill's list.

    Yes: --min-volume (0 to 100000) and --max-difficulty (0 to 100). The score is fixed: search volume / (keyword difficulty + 1).

    New in Marketing & SEO