# CrazySEO [Health: Active]

**Category:** 🎯 Marketing  
**Repository:** https://github.com/d08m/crazyseo  
**GitHub Stars:** 4  
**npm Downloads (last month):** 439  
**Views:** 2  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/crazyseo-2

## Description
Audit whether ChatGPT, Gemini and Perplexity recommend a site. Readiness checks need no keys.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "crazyseo": {
    "command": "npx",
    "args": ["-y","crazyseo"]
  }
}
```

## Documentation

## What CrazySEO MCP server does

CrazySEO MCP server evaluates how likely a website is to appear in AI-generated recommendations. It separates technical readiness from model visibility: readiness checks inspect the site directly, while visibility audits ask buyer-style questions without naming the target brand.

The project can report whether configured engines mention the site, which competing brands appear instead, and which third-party sources are cited on questions the site loses. Results retain the question and raw model response so findings can be checked manually.

## How it works

A readiness audit fetches the target site and runs 11 checks covering HTTPS, crawler access in robots.txt, schema.org markup, title and meta description, JavaScript-independent readability, canonical URLs, Open Graph tags, H1 structure, image alt text, sitemap references, and sampled internal links. `llms.txt` is displayed as additional information but is not scored.

For visibility testing, the software reads the site, derives what it offers, creates questions likely to be asked by buyers, and sends those questions to configured engines. The target brand is excluded from the questions to avoid forcing a mention. If an engine fails or is rate-limited, the report identifies the missing coverage rather than treating the remaining results as complete.

## Setup and configuration

The readiness command needs no account, API key, or configuration:

```bash
npx crazyseo readiness example.com
```

Source installation is also supported through Git, npm, and the project build command. JSON output is available for automation, and readiness exits with a non-zero status when checks fail, allowing it to gate a deployment.

Cross-engine visibility requires whichever provider keys are available: `OPENAI_API_KEY`, `GOOGLE_API_KEY`, and `PERPLEXITY_API_KEY`. Any one key can be used, with additional keys increasing engine coverage. The MCP server can be registered over stdio with the `crazyseo mcp` command. A Claude Code plugin is also provided.

## Tools and capabilities

The CrazySEO MCP server exposes these tools:

- `readiness_check`: Runs the 11-point site audit without keys.
- `audit_plan`: Produces site facts and unbranded buyer questions.
- `audit_score`: Scores supplied findings for visibility, rankings, and citation gaps.
- `audit_visibility`: Runs blind visibility checks across configured engines and preserves answers.
- `citation_gap`: Finds sources cited on questions the site does not win.
- `engine_status`: Reports which engines the installation can query.

The package can also be used as a library through its `checkReadiness` function, which returns a score and individual checks.

## Limitations and notes

Keyless planning and scoring can use the agent's own site access and web-search capabilities, but cross-engine comparisons require provider keys. Readiness is not the same as visibility, and a score based on only some engines must be interpreted as partial coverage. The audit declines to score an unreachable site rather than estimating a result. The project is self-hosted, provides no telemetry or account system, and is licensed under AGPL-3.0.

_Full upstream README: https://allmcps.com/mcp/crazyseo-2/readme_