Friends beta

Make everything your AI writes sound human.

Humanizer is an MCP server. Plug it into the AI tools you already use. It finds the words and patterns that make text sound machine-made, then tells your model exactly how to fix them.

Before (score 30)

In today's fast-paced digital landscape, it's important to note that leveraging cutting-edge solutions can truly unlock unparalleled value. Moreover, we delve into a myriad of transformative strategies.

After (score 100)

Online business moves fast. New tools can add real value. Here we cover a few strategies that make a large difference.

How it works

check_text

Scores any draft from 0 to 100. Lists every banned word, filler phrase and robotic pattern, with the exact spot and a fix.

humanize_text

Does the safe fixes for you, then hands your model short rewrite instructions and a style guide. Your model rewrites. Then it checks again.

Your model, no extra AI bill

The server never calls an AI model. The rewrite runs on the model you already pay for. Checks are instant and the same every time.

Works where you work

Cursor, Claude Code, Claude Desktop, OpenClaw, Hermes, and any client that speaks MCP over Streamable HTTP. One URL and one key. Setup takes about a minute.

What it catches

Pricing

$1first month

Then $5/month. Unlimited use across all your AI tools. Cancel anytime.

  • Both tools, every client
  • 60 calls per minute per key
  • Up to 20,000 characters per call
Coming soon. Paid plans are not live yet. Right now it's a free friends beta.
Get a key

Questions

Does it send my text to another AI?

No. The server runs fixed rules on your text and returns a score and instructions. It does not call any AI model.

Will my model actually follow the fixes?

The instructions are short and specific to each problem. Your model rewrites, then calls check_text again to confirm.

Do I need to install anything?

Not for most tools. Paste a URL and a key into your MCP config. Claude Desktop needs a small bridge that runs with npx.