# Not Human Search [Health: Active]

**Category:** 🔎 Search & Data Extraction  
**Repository:** https://github.com/unitedideas/nothumansearch-mcp  
**GitHub Stars:** 13  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/not-human-search-2

## Description
Search the agentic web. 960+ sites scored 0-100 on agentic readiness. Includes verify_mcp tool.

## Tools
Capabilities this server exposes over MCP:

- **search_agents** — Find agent-ready websites, APIs, and services by keyword, category, or minimum readiness score.
- **get_site_details** — Get the full 7-signal readiness report for a specific domain.
- **get_stats** — Index statistics: total sites, average score, top category.

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "not-human-search": {
    "url": "https://nothumansearch.ai/mcp"
  }
}
```

## Documentation

## What Not Human Search MCP server does

Not Human Search MCP server provides an MCP interface for discovering web resources designed for interaction by AI agents. Its index contains more than 960 sites, including websites, APIs, and services. Search results can be filtered by keyword, category, or minimum agentic-readiness score.

The service is intended for discovery rather than general web browsing. It helps an agent identify candidate resources and then inspect the signals used to evaluate a particular domain. A separate website, REST API, on-demand checking endpoint, badge endpoint, and submission endpoint are also listed, but the MCP connection exposes the tools described below.

## How it works

The server is hosted at `https://nothumansearch.ai/mcp` and uses streamable HTTP transport. An MCP-compatible client connects to that URL and invokes the available tools. No local package installation is required for the hosted endpoint.

Readiness scores range from 0 to 100 and are based on seven signals:

- `llms.txt`
- `ai-plugin.json`
- An OpenAPI specification
- A structured API
- An MCP server
- AI bot rules in `robots.txt`
- Schema.org markup

The signals have different weights. `llms.txt` contributes up to 25 points, while Schema.org markup contributes up to 5 points. The complete weighting is available through the project documentation and is reflected in the reported site details.

## Tools and capabilities

- `search_agents` finds agent-ready websites, APIs, and services using a keyword, category, or minimum score.
- `get_site_details` returns the seven-signal readiness report for a specified domain.
- `get_stats` reports index totals, average score, and the top category.

The Not Human Search MCP server fits workflows that need to select external services before attempting an agent-driven integration. For example, an agent could search for an email API with a minimum readiness score, compare candidate domains, and inspect each domain's signals before choosing one.

## Setup and configuration

Connect an MCP client to the hosted endpoint. The README gives this Claude command:

```bash
claude mcp add --transport http nothumansearch https://nothumansearch.ai/mcp
```

The same endpoint is described as usable by MCP-compatible clients, including Cline, Cursor, Zed, Goose, and Claude Code. The provided material does not document environment variables, API keys, client-specific configuration files, or local runtime requirements.

## Limitations and notes

The readiness score measures the presence of seven published signals; it does not establish that a service is reliable, available, safe, or suitable for a particular production workload. A high score should therefore be treated as a discovery signal rather than a guarantee of successful integration.

The MCP server source is described as private. The README itself is MIT-licensed. The material does not state pricing, rate limits, uptime commitments, authentication requirements, or how frequently the index is refreshed. The separate REST endpoints should not be assumed to be MCP tools.

Not Human Search MCP server is most appropriate when an agent needs a ranked starting point for finding callable resources, not when it needs to execute arbitrary actions on the indexed sites.

_Full upstream README: https://allmcps.com/mcp/not-human-search-2/readme_