Skip to main content
Glama
AI2Human MCP Server

AI2Human MCP Server

Official
by ai2humanapp

AI2Human MCP Server

Humanize AI-written text and detect AI writing, straight from Claude, ChatGPT, Cursor and any other MCP client.

MCP Auth Hosted License

Website · Setup guide · Pricing · Tools · Support


AI2Human is a hosted Model Context Protocol server. Connect it once and your AI agent can rewrite machine-sounding text so it reads as natural human writing, check how AI-like a passage looks, and report your remaining allowance. There is nothing to install or run: you add one URL.

https://ai2human.app/api/mcp

Transport

Streamable HTTP

Sign-in

OAuth 2.1 with PKCE and dynamic client registration, or an API key

Access

Paid plans: Basic, Pro and Ultimate

Hosting

Remote, hosted by AI2Human. No local process

About this repository. It holds the public documentation, the MCP Registry metadata (server.json) and an issue tracker for the hosted server. The service itself is proprietary and is not open source.

Tools

Tool

What it does

Cost

humanize_text

Rewrites AI-generated text as natural human writing while keeping meaning, facts, figures and citations. Eight styles.

1 word per input word

detect_ai

Estimates how much of a text reads as AI-generated, with a per-sentence breakdown.

1 detector scan

get_usage

Shows your plan, words used and remaining, the per-call word limit and detector scans used. Read-only.

Free

Every tool carries MCP annotations (readOnlyHint, destructiveHint, idempotentHint, openWorldHint). None of them deletes or changes anything beyond your own usage counters. Full parameters, outputs and examples are in docs/tools.md.

Related MCP server: Natural Voice MCP

Quick start

  1. Sign in to an AI2Human account on a paid plan (pricing).

  2. Add the server URL to your AI app. Per-app steps are below and in docs/clients.md.

  3. Approve the connection when AI2Human asks, or paste an API key.

  4. Ask your agent: "Humanize this paragraph with AI2Human."

Claude (claude.ai and desktop)

Settings → Connectors → Add custom connector → paste https://ai2human.app/api/mcp → Connect → approve on the ai2human.app page. No API key needed.

Claude Code

claude mcp add --transport http ai2human https://ai2human.app/api/mcp

Then run /mcp inside Claude Code and choose ai2human to sign in through the browser. To use an API key instead:

claude mcp add --transport http ai2human https://ai2human.app/api/mcp \
  --header "Authorization: Bearer YOUR_API_KEY"

Cursor, VS Code and other clients

Add this to the client's MCP config (for Cursor, .cursor/mcp.json):

{
  "mcpServers": {
    "ai2human": {
      "url": "https://ai2human.app/api/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}

Clients that support OAuth can leave out headers and sign in through the browser. Clients that only support local (stdio) servers can use the mcp-remote bridge with the same URL.

ChatGPT

In ChatGPT settings, open connectors, add a custom connector (MCP server), paste the URL and choose OAuth. Custom connectors depend on your ChatGPT plan and may need developer mode switched on.

Example

You: Check this paragraph with AI2Human. If it reads as AI, humanize it in an academic style and tell me the new score.

The agent calls get_usage, then detect_ai, then humanize_text with style: "academic", and returns the rewritten text with a line such as:

[AI2Human <engine>] Words charged: 214. Words remaining: 49786. Our detector scores the result 94% human.

Plans and limits

MCP uses the same allowance as the website. There is no separate MCP price. Figures below are current at the time of writing; the pricing page is the source of truth.

Plan

Words per month

Words per call

Basic

15,000

600

Pro

50,000

2,000

Ultimate

100,000

5,000

Each account can make 30 tool calls per minute. To humanize text longer than the per-call limit, split it at paragraph breaks and call humanize_text once per part. If a call fails, no words are charged.

Security and privacy

  • A key or connection can call only the three tools above. It cannot read your password, billing details or account settings.

  • API keys and OAuth tokens are stored only as hashes. A key is shown once, when you create it.

  • Connected apps sign in with OAuth 2.1 and PKCE. Access tokens last one hour and renew automatically. A connection expires 90 days after approval and must then be approved again.

  • You can disconnect an app or revoke a key at any time in Dashboard → Settings, and it stops working immediately.

  • Text sent through MCP is handled like text entered on the website. See the privacy policy.

To report a vulnerability, see SECURITY.md.

Discovery metadata

For tooling that discovers servers automatically:

Troubleshooting

See docs/troubleshooting.md for connection problems and every error code.

Support

License

The documentation and configuration in this repository are released under the MIT License. The hosted AI2Human service is proprietary and covered by the Terms of Service.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      Not graded
      maintenance
      Provides tools and resources to detect AI-generated writing patterns and refine text for more authentic, human-like communication. It enables users to analyze phrasing via a scoring system and apply conversational guides tailored for platforms like Twitter and LinkedIn.
      1
      -
    • A
      license
      A
      quality
      B
      maintenance
      Enables AI assistants to rewrite AI-styled English prose into more natural human writing, with tools for humanizing text and retrieving information about the underlying model.
      2
      133 npm
      1
      Apache 2.0
    • F
      license
      Not graded
      quality
      A
      maintenance
      Enables agents to analyze prose for formulaic AI writing patterns and receive a 0–100 score, band label, contextual violations, and concrete advice through tools for inline text or files.
      -