Open source · Local-first · MCP server

Make AI sound like you.

Your agent drafts the email, the PR, the Slack reply. HumanifyMe rewrites it in your actual voice, learned from your real writing, stored only on your machine.

Free & open source · Bring your own API key · No account

See it in action

One command sets up your profile. Then rewrite any draft in your voice.

humanifyme-demo

$

A voice fingerprint, not a filter

“AI humanizers” rewrite toward a generic professional register. HumanifyMe builds a structured profile of how one specific person writes.

01

Give it your writing

Paste 3–10 real emails or messages, point it at a folder, or import your ChatGPT / Claude history. Emails, phone numbers and keys are masked before any model sees them.

02

It learns your fingerprint

Not an opaque embedding — a readable, editable profile: sentence rhythm, formality, your actual phrases, words you avoid, how you disagree.

03

Your agent uses it

Every MCP agent gets a humanify_text tool. Ask it to humanify a draft, or let the bundled skills do it after your agent writes an email or PR.

Install in one command

Requires Node 22.5+. Register it with each agent you use; they all share one voice profile.

Claude Code

Plugin: MCP server plus three skills.

claude plugin marketplace add JoshMcQ/HumanifyMe claude plugin install humanifyme@humanifyme

Cursor, Claude Desktop, etc.

Add this server to your mcp.json.

{ "mcpServers": { "humanifyme": { "command": "npx", "args": ["-y", "--package", "humanifyme@latest", "humanifyme-mcp"] } } }

Command line

Full pipeline without an agent.

npx -y humanifyme@latest setup npx -y humanifyme@latest rewrite draft.txt

No API key? Analyze first.

Local, deterministic AI-writing checklist.

echo "At its core, this robust approach paves the way." | npx -y humanifyme@latest analyze

Help us grow

HumanifyMe is Apache 2.0 and built in the open. Stars and contributors make the project visible and keep it honest.

🐛 Issues 💬 Discussions

Fair questions

Is this an “AI detector bypass”?

No. HumanifyMe makes text sound like you because it is built from how you write, not from tricks aimed at fooling a classifier.

What do you collect about me?

No account, and your samples and profile stay in ~/.humanifyme/ on your machine. What leaves is redacted text sent to the LLM provider you choose: your samples once, to build the profile, then each draft you rewrite. (The browser trial is different: it sends text to Claude from our server and stores nothing unless you tick "share".) Setup offers to share anonymous quality counts (never text); it is off unless you opt in.

Does it work with local models?

Yes. Point it at Ollama and nothing ever leaves your machine, not even redacted text.