The AI content checker that runs before you publish.
Plagiarism, fact-check, tone, legal risk, and 6 more quality checks — in parallel, in 30 seconds. Open source.
I ghostwrote 200+ articles using AI last year. Every. Single. One. had a plagiarism hit, a legal risk, or a broken fact.
So I built this. Before you hit publish, every article runs 9 checks in parallel in under 30 seconds. Around $0.15 per article (estimate — varies by provider). Open source. MIT.
— Sharon Sciammas, creator
9 skills. 30 seconds. One verdict.
Here's what a real check looks like. Paste an article, get a per-skill score and an overall verdict.
9 quality checks. One command.
Every skill runs in parallel. Enable only what you need — disable the rest. Bring your own API keys.
Plagiarism
Checks the indexed web. Returns similarity % + matched sources.
AI Detection
Detects AI-generated passages. Per-sentence probability + verdict.
SEO
Word count, headings, readability, keyword density. No API cost.
Fact Check
Extracts claims, searches evidence, rates each with confidence.
Tone of Voice
Compares against your brand voice guide. Flags violations with rewrites.
Legal Risk
Scans for health claims, defamation, GDPR, false promises.
Content Summary
Topic, argument, audience, tone classification.
Brief Matching
Checks article against your content brief requirements.
Content Purpose
Classifies article type with purpose-specific recommendations.
Running in 60 seconds
From zero to your first report.
Install
npm install -g checkit
One command with npm. Or download a binary from GitHub Releases — Mac, Linux, Windows.
Configure
checkit --setup
Interactive wizard. Paste your API keys for Copyscape, MiniMax, Exa. Skills activate as keys are added.
Run
checkit ./my-article.md
Report opens in your browser in under 30 seconds. HTML, JSON, or Markdown — your pick.
Built for your workflow
Use it the way you already work.
CLI
One command from your terminal. Outputs HTML, JSON, or Markdown.
Dashboard
Local web UI for browsing history, toggling skills, managing contexts.
MCP Server
8 tools for Claude Code, Cursor, Windsurf. Agents call check_article directly.
CI Mode
--ci flag exits 1 on fail. Gate pull requests before merging content.
JSON API
Structured output for scripts. Pipe to anything.
Custom Skills
Write your own validators in TypeScript. One interface to implement.
One dashboard. All your checks.
Runs locally with checkit --ui. Your data never leaves your machine.
Every check, every verdict, every score
Browse your entire check history. Filter by verdict, search by source, click into any report for the full breakdown.
Toggle on, toggle off
Enable only what you need. Each skill shows its engine label and API key status so you always know what's ready.
Brand voice, legal policy, briefs
Upload your tone guide once. CheckApp applies it to every article going forward. Same for legal policies and content briefs.
Bring your own LLM
Pick MiniMax, Anthropic, or OpenRouter. Swap providers anytime. CheckApp charges nothing — you pay the provider directly.
Every check, every verdict, every score
Browse your entire check history. Filter by verdict, search by source, click into any report for the full breakdown.
Who uses it
People who write. Whatever their title.
Gate every article before publish
Connect CheckApp to your content pipeline. Block any article scoring below your threshold — no more brand damage from AI slop.
Quality control at scale
Batch-check every client article. Track scores over time in the dashboard. Flag problems before the client sees them.
Verify your own writing
Running a newsletter, blog, or SEO site? Every article gets plagiarism, fact-check, and tone review in under 30 seconds.
Integrate with your agents
MCP tools for Claude Code and Cursor. --ci flag for PRs. JSON API for scripts. Bring your own AI agent.
What it costs to run
The tool is free. You pay the API providers directly. Here's the full breakdown.
- MIT-licensed open source
- No platform fee. Ever.
- Zero telemetry. Zero tracking.
- Fork it, modify it, resell it.
* Bring your own API keys. CheckApp charges nothing.
Prices above are estimates based on public provider pricing at time of writing, for an 800-word article checked with MiniMax as the LLM. Your actual cost depends on article length, which LLM provider/model you choose (MiniMax, Anthropic, OpenRouter, etc.), and vendor price changes. Longer articles and premium models (GPT-4, Claude Opus) will cost more. Use a cheaper model or disable skills you don't need to lower the bill.
Powered by best-in-class APIs
We pick the best tool for each job. Swap any of them via environment variable.
Sharon Sciammas — builder, based in Amsterdam.
Built 3 startups (raised $1M). Co-founded W3B Lab — Europe's largest blockchain community (8 cities, 4K members). Most recently led AI transformation at a multi-product company — 4× output, 70% cost reduction.
CheckApp is the content quality tool I needed most. It's MIT-licensed — fork it, break it, rebuild it.
From the blog
Notes on AI content, quality gates, and building open-source tools.