Published in the official MCP Registry

An SEO MCP server your agent can call mid-workflow.

Connect Claude Code, Codex, Cursor, or another MCP client to 45 live tools for SERP research, content, backlinks, local SEO, rank tracking, and AI visibility.

Get Free API Key Choose an install path

45
discoverable tools: 2
connection paths: 1
workflow contract

agentseo.mcp

Hosted MCP

claude mcp add --transport http agentseo https://www.agentseo.dev/mcp \
  --header "x-api-key: sk_live_your_key" \
  --header "x-project-id: client-alpha" \
  --header "x-workflow-id: nightly-refresh"

First useful prompt

Use AgentSEO to find content gaps for example.com around "seo api for ai agents". Then turn the strongest supported opportunity into a content brief with a title, outline, and evidence.

Choose the transport

Hosted for less setup. Local for client-level control.

Both paths expose the same AgentSEO tool catalog. Choose based on where your MCP client runs and how you manage server-side secrets.

Hosted Streamable HTTP

Use when your client supports remote MCP and you do not want a local wrapper.

claude mcp add --transport http agentseo https://www.agentseo.dev/mcp \
  --header "x-api-key: sk_live_your_key" \
  --header "x-project-id: client-alpha" \
  --header "x-workflow-id: nightly-refresh"

Local npm stdio

Use for desktop clients or when the MCP process should run beside your repository.

{
  "mcpServers": {
    "agentseo": {
      "command": "npx",
      "args": ["-y", "@agentseo/mcp-server"],
      "env": {
        "AGENTSEO_API_KEY": "sk_live_your_key",
        "AGENTSEO_API_URL": "https://www.agentseo.dev/api/v1"
      }
    }
  }
}

Keep API keys server-side. Use separate project or workflow identifiers when you need attribution across clients, tenants, or automations.

Use it inside the work

The model can inspect live evidence before it writes or recommends.

MCP is useful when the caller is already an agent. It can discover the relevant tool, supply the inputs, and keep the result inside the working context.

Inspect workflow bundles

SERP and keyword research

Give the model live search context instead of asking it to infer demand and competition from stale training data.

Content planning and QA

Move from content gaps to outlines, briefs, schema plans, internal links, and pre-publish checks.

Backlinks and competitors

Inspect backlink summaries, referring domains, competitor intersections, and qualified opportunity inputs.

Local and rank tracking

Run local audits, local visibility checks, rank tracking, and job polling inside the same agent loop.

AI search visibility

Build stable prompt sets and inspect AI Overview results with repeatable inputs instead of ad hoc screenshots.

Attributed execution

Keep discovery public while tool execution uses a server-side key plus optional project and workflow attribution.

Public verification

Inspect the listing, not just the claim.

Registry metadata, the detected tool catalog, and the installable package are available from independent public surfaces.

Official MCP Registry

Active listing · registry version 0.1.4 · hosted and npm metadata

Smithery

Live listing · 45 tools exposed for inspection

npm

@agentseo/mcp-server · package version 0.1.3

Choose MCP when

The model should discover and call the tool.

Useful for coding agents, interactive research, writing, debugging, and operator-reviewed workflows.

Choose REST when

Your backend should control every request.

Useful for deterministic services, queues, scheduled jobs, and integrations with explicit endpoint handling.

Before the first tool call

The setup questions that change the decision.

Is AgentSEO hosted or a local npm MCP package?

Both. Use the hosted Streamable HTTP endpoint when you do not want a local wrapper. Use the npm package for local stdio clients and repository-level configuration.

Does it work with Claude Code, Codex, and Cursor?

AgentSEO works with MCP-capable clients that support hosted HTTP or local stdio servers. The exact setup depends on the transport your client supports.

Do I need an API key?

Yes for tool execution. Discovery calls can inspect the server and tool catalog, while paid execution requires an AgentSEO server-side API key.

What should I run first?

Start with one bounded job such as a content-gap check or local audit. Review the evidence and output shape before you automate a larger workflow.

When should I use REST instead?

Use REST for a direct backend integration with explicit endpoint control. Use MCP when the model should discover and call SEO tools while it plans, writes, debugs, or monitors.

Start with one real job

Connect one workflow before you automate ten.

Create a server-side key, choose hosted or local, and test one bounded workflow with inputs you can review.

Get Free API Key Read the Quickstart