A one-writer version of Clearscope is a pipeline: take a query, fetch the top-ranking pages, pull out the terms and headings they share and score a draft on how many it covers. An agent builds that in a weekend with a SERP API and a text extractor. What it can't copy is trust in the grade: Clearscope tunes its recommendations on its own language analysis and search data, and a letter only helps if it tracks rankings. You also depend on a paid SERP API for fresh results, and tracking brand mentions in ChatGPT and Gemini adds more outside services.
Before you buildClearscope has no free plan, only a 14-day trial. With one or two articles a month, exporting the term CSV during the trial may be all you need.
Build me a content grader for my own articles to replace Clearscope. Requirements: - Node.js 22 + TypeScript: a Fastify server and a React editor built with Vite. `npm start` serves http://localhost:4720. - SERP source: Serper.dev (SERPER_API_KEY in .env) for the top 20 organic results of a query, with country and language chosen per report. Cache every response in SQLite for 7 days. - Page text: fetch each result with `undici`, extract the main content with `@mozilla/readability` and `jsdom`, skip PDFs and pages that still fail after one retry, and keep headings, word count and text. Obey robots.txt and wait 1 second per host. - Storage: better-sqlite3 in ./data/grader.db for queries, fetched pages, term lists, drafts and every grade over time. - Terms: extract 1 to 3 word phrases with `wink-nlp`, drop stop words and rank by the share of results that use each phrase, weighted by inverse document frequency over every page fetched so far. Show the top 60 with that share and the typical count per page. - Editor: paste or write a draft; each term shows as used, missing or overused (above the 90th percentile of the competitors), with the word count against the competitors' median and a letter grade A to F from the weighted share of terms covered. Re-grade 1 second after typing stops. - Export: the term list as CSV, the draft as Markdown or HTML, and a read-only report page I can send to a writer. - Single user. No accounts, no telemetry, local-first: only SERP API calls and page fetches leave the machine. - Out of scope: Google Docs and WordPress plugins, a content inventory, chatbot tracking, seats. - README: setup, the SERP API cost of one report, and a warning that the grade is a heuristic to compare with your own ranking results before trusting it.
Opens your agent with the prompt filled in, you press enter. Or copy it and paste it into an empty folder.
The first prompt builds the core. Run these one at a time in the same folder to grow it into your own Clearscope.
Add a decay check to the content grader in this folder. Requirements: - Pull clicks and impressions per page and query for the last 90 days through the Search Console API with a service account (path in .env). - Flag pages whose clicks fell more than 30% against the previous 90 days, with their top query. - For each flagged page, fetch it, re-run the grader against fresh results for that query and show the old and new grade. - Output a CSV and a page in the UI. Update the README with the service-account steps.
Add an outline generator to the content grader in this folder. Requirements: - From the stored headings of the top 10 results, cluster similar headings (normalised text plus token overlap) and list the sections that 4 or more results share. - With an OpenAI-compatible endpoint from .env, turn that list into a suggested H2/H3 outline and show which competitor headings each section came from. - Insert the outline into a new draft; never insert sentences from competitor pages. - Update the README.
Add a page comparison to the content grader in this folder. Requirements: - Given a URL and a query, fetch the page, grade it and list the missing and overused terms against the top results, excluding the page's own domain. - Show word count, heading count and the top 10 missing terms next to the top 3 competitors. - Export the comparison as Markdown. - Update the README.
Editors get one letter grade they can hold a writer to, and a term list built from real results instead of a guess. Unlimited seats make it easy to hand to freelancers, and the price is small next to an article that never ranks.
Estimate · Runs on your machine; SERP lookups for a handful of reports a month cost a few dollars at most. The build is token usage on your agent plan.
| Plan | Monthly | Annual, per month | What you get |
|---|---|---|---|
| Essentials | $129 | — | 50 tracked queries, 50 pages, 20 topic explorations and 20 drafts a month, unlimited users and projects. |
| Business | $399 | — | 300 tracked queries, 300 pages, 50 topic explorations and 20 drafts a month, unlimited users and projects, a dedicated account manager. |
| Enterprise | custom | — | Everything in Business plus custom credits and agreements, crawler whitelisting and single sign-on. |
No free plan; a 14-day free trial.
Kinda. The core is a weekend build with the prompt on this page, but real gaps remain: Recommendations and a grade tuned on Clearscope's own search and language data, Content Inventory that re-scores published pages and flags decay from Search Console data.
About $129 a month on the Essentials plan, $1,548 a year (checked 2026-10-19).
Recommendations and a grade tuned on Clearscope's own search and language data; Content Inventory that re-scores published pages and flags decay from Search Console data; The Google Docs add-on and WordPress plugin that put the grade where writers draft; Tracking of when and how your brand shows up in ChatGPT and Gemini answers; Unlimited seats, onboarding and support included in the price. If any of those matter to you, keep paying.
Yes: BlogPilot AI (open source).