OctoBoost SEO — MCP Server

Exposes the OctoBoost SEO API as Model Context Protocol (MCP) tools so AI agents can audit websites without leaving their reasoning loop.

Get your free API key at octo-boost.com — new accounts come with free credits to try out every tool.

Why use this for SEO analysis?

Running SEO checks inside an LLM context is expensive — fetching raw HTML, parsing it, and reasoning over it consumes thousands of tokens per page. This server offloads that work to the OctoBoost API and returns compact, structured results (scores, flags, and diagnostics only) that cost a fraction of the tokens while giving the agent exactly what it needs to reason and act.

• Token-efficient — structured API results instead of raw HTML; a full 30-analyzer audit of a page fits in a few hundred tokens.
• LLM-friendly output — scores and pass/fail flags, not prose. No parsing required.
• Scoped execution — run one analyzer, one category, or a full audit. The agent picks the right scope and avoids unnecessary API calls.
• Real-time progress — analyze emits a notifications/progress event after each URL, so the agent and user get live feedback.
• Credit awareness — every response includes credits used/remaining so the agent can make cost-conscious decisions.
• Clear error signals — 401 for expired keys, 402 for exhausted credits. Easy for an agent to catch and report.

Tools

list_analyzers

Returns all available analyzer keys, their categories, and their weights.

Call this at the start of any audit session to let the agent know what it can work with.

Current analyzer keys:

• alt-tags
• color-contrast
• title
• meta-description
• heading-hierarchy
• canonical
• aria-attribute-check
• error-handling-check
• form-check
• h1-check
• hreflang
• html-lang
• html-structure
• image-optimization
• image-size
• internal-links
• link-descriptions
• list-markup
• lorem-ipsum
• meta-viewport
• robots-meta
• robots-txt
• script-loading
• script-size
• social-meta
• structured-data
• tab-focus-order
• touch-target-size
• url-structure
• xml-sitemap

No input required.

Found 30 analyzers across 4 categories.

Categories: seo, accessibility, ux, performance

Analyzers:
  title (categories: seo)
  meta_description (categories: seo)
  ...

---

scan_domain

Crawls a domain and returns all SEO-relevant page URLs.

Parameter	Type	Default	Description
domain	string	—	Domain or URL to scan (e.g. example.com)
maxPages	number	100	Max pages to crawl (max: 500)
excludePatterns	string[]	[]	URL patterns to skip
respectRobotsTxt	boolean	true	Honour robots.txt
defaultLanguageOnly	boolean	true	Skip alternate-language duplicates

---

analyze

Analyzes one or more URLs for SEO issues. URLs are processed sequentially with progress notifications emitted after each.

Parameter	Type	Description
urls	string[]	URLs to analyze (max: 20)
mode	"full" | "analyzer" | "category"	Scope of analysis
analyzer	string?	Required when mode is "analyzer"
category	string?	Required when mode is "category"

Mode reference:

Mode	What runs	Typical use
full	All 30 analyzers	Comprehensive site audit
category	All analyzers in one category	Focused audit (e.g. accessibility)
analyzer	One specific analyzer	Targeted check (e.g. title)

Setup

1. Get an API key — sign up at octo-boost.com. New accounts get free credits.

2. Add to your MCP client — no install needed, npx runs it automatically.

{
  "mcpServers": {
    "octoboost-seo": {
      "command": "npx",
      "args": ["-y", "octoboost-mcp-server"],
      "env": {
        "OCTOBOOST_API_KEY": "your-api-key"
      }
    }
  }
}

Add this block to your client's MCP config file:

• Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json
• Cursor: Cursor MCP settings file
• OpenClaw: ~/.openclaw/mcp.json

3. (Optional) Verify — run the MCP Inspector and call list_analyzers:

OCTOBOOST_API_KEY=your-key npx @modelcontextprotocol/inspector npx octoboost-mcp-server

Example agent workflow

1. list_analyzers
   → agent learns categories and picks "accessibility" for the task

2. scan_domain { domain: "acme.com", maxPages: 50 }
   → agent receives 38 URLs

3. analyze { urls: [first 10 URLs], mode: "category", category: "accessibility" }
   → progress: 1/10 … 10/10
   → agent reads scores, identifies 3 failing pages

4. analyze { urls: [3 failing URLs], mode: "full" }
   → agent gets the complete picture and writes a remediation report