Does your writing sound like AI? Paste it. Every AI-writing tell gets highlighted with a fix. Runs in your browser, nothing is uploaded.
20+ patterns that make text read as machine-written: "It's not X, it's Y" contrasts, dramatic one-line closers, rows of fragments, faux-insight setups ("what nobody tells you"), staged run-ups ("let's dive in"), inflated significance ("a pivotal moment", "a testament to"), shallow -ing riders, sales language, borrowed authority ("experts agree"), AI vocabulary (delve, tapestry, seamless, leverage), em-dash overuse, bold-label lists, emoji bullets, and chatbot leftovers ("I hope this helps!"). Single weak signals (one dash, one triad) don't count against you unless other tells are present.
Read the guides: "It's not X, it's Y" · "delve" and other AI words · em dash overuse · rule-of-three lists · hedging openers · all patterns.
Unlimited checks in the browser. API up to 5,000 characters per call.
For writers and teams: an API key that lifts the free 5,000-character limit to 200,000 characters per check, for long documents and your own scripts. Works with the open-source CLI and GitHub Action. Email support from the person who built it.
What's includedNo account. Pay per call in USDC over x402. Long texts up to 200k chars, batches of 50 for $0.02. MCP server included.
For agencies that ship client copy: batch checks on client drafts, a per-client report you can send with each delivery, saved history, and API keys for your own tools. The founding price is for the first 25 agencies and stays at $49/mo for 12 months.
Checkout isn't open yet, and nothing is charged now. To hold a founding spot, email slopscore@agentmail.to with the word "founding", or join the waitlist:
We'll email you once when checkout opens. You decide then.
402 recent posts from 81 agency blogs, scored and anonymized. 22% scored 20 or higher.
Read the study
Free AI content policy template
Methodology and limits
Agency plan details
curl -s -X POST https://slopscore-nine.vercel.app/api/check -H 'content-type: application/json' \
-d '{"text":"Great question! Let\u0027s dive in. This isn\u0027t just a tool, it\u0027s a mirror."}'
# long texts / batches (x402, USDC on Base): POST /api/paid/check, POST /api/paid/batch
# MCP (streamable-http): https://slopscore-nine.vercel.app/mcp tools: slop_check, slop_check_long, slop_patterns
# CLI for CI: npx github:kburrus64-max/slopscore README.md docs/*.md --max 30
Reviewing AI-written pull requests? Try SlopScore for Code, the PR gate for AI slop in diffs.
Docs for agents: /llms.txt · pattern list · x402 discovery. SlopScore flags writing patterns; it does not claim to prove who or what wrote a text. Pattern taxonomy credits: blader/humanizer and petergyang/no-ai-slop (MIT), building on Wikipedia's "Signs of AI writing".
Methodology · Privacy · Terms · Contact: slopscore@agentmail.to