Why not call an upstream provider API directly?

You can, but you’ll still build a job model, normalize payloads, and keep response shapes stable for tool use. AgentSEO wraps that into strict JSON + markdown_summary outputs designed for agents.

Do responses include markdown, or only JSON?

Core endpoints return structured JSON, and key workflows include a markdown_summary field so you can share or store a human-readable brief alongside the data.

Sync vs async: how does it work?

Use ?sync=true to block briefly and get the result when possible. Otherwise you’ll receive a jobId (202) and poll /api/v1/jobs/{id} until it completes.

Do I need a credit card to try it?

No. Hobby includes 100 one-time live credits, plus mock mode for local development and demos.

How should I use this safely before automating changes?

Treat outputs as decision support: keep sources, set thresholds, and require human approval before publishing or making high-impact edits. Start with weekly checks, then automate only what you can validate.

100 free credits. No credit card.

SEO API your agentscan actually use

Name: AgentSEO
Brand: AgentSEO

Make one request, get compact SEO intelligence your agent can actually use, then move the same contract into REST, MCP, or your backend workflow.

Create key -> Run first request -> Reuse the workflow in your runtime

Try Live Demo Read Docs

Deterministic JSONMCP + RESTType-safe inputs

Deterministic JSONMCP + RESTType-Safe Inputs

agentseo-run.ts

import { AgentSEO } from "@agentseo/sdk";

const client = new AgentSEO({
  apiKey: process.env.AGENTSEO_API_KEY!
});

const result = await client.localVisibilityTrack(
  {
    domain: "northbeamcoffee.com",
    keywords: ["coffee near me"],
    locations: ["Austin, TX"]
  },
  { sync: true }
);

console.log(result);

Response

Compact, LLM-ready output

Execution

Sync + async job modes

Coverage

14 ready-to-run workflows

Start Here

Pick the job first, then take the shortest path

This is the fastest way through the site for the three ICPs we care about most. Choose the path that matches the work you need to do next.

Developers

Make one successful request

Start with quickstart, inspect the response contract, then move into the API reference once the first call works.

Open quickstart

Growth Engineers

Pick the workflow bundle

Start with packaged ICP playbooks for launch, refresh, AI visibility, programmatic SEO, and competitor-gap work.

Open workflows

Agencies

Find the client-facing workflows

Go to pricing and integration paths if the next question is how to package repeated audits, monitoring, or content-gap work.

Review pricing

Proof Paths

Validate fit with one real workflow, not a generic demo

The fastest way to trust AgentSEO is to match one live problem to one concrete guide. These are the three proof-heavy paths most likely to move an ICP visitor from curiosity into a first successful request.

Make first request

API buyer proof

Compare API shape before you compare endpoint count

Best for developers and growth engineers deciding whether they need a raw provider, a SERP API, or a workflow-shaped layer.

Workflow proof

See what a safe SEO agent loop actually looks like

Best for builder-marketers and product engineers who need to separate collection, decision, and action before anything touches production.

Measurement proof

Build the AI visibility loop before you build the dashboard

Best for agencies and operators who need a repeatable prompt, citation, and reporting system instead of screenshot theater.

Works with your agent stack

Works for no-code, low-code, and developer workflows

REST API

MCP Server

LangChain

n8n

Codex

OpenClaw

Claude Desktop

OpenAI Functions

Vercel AI SDK

Custom Agents

REST API

MCP Server

LangChain

n8n

Codex

OpenClaw

Claude Desktop

OpenAI Functions

Vercel AI SDK

Custom Agents

Integration Guides

Pick the runtime, then copy the shortest proven path

Use runtime-specific guides when the packaging matters. Some paths are MCP, some are HTTP workflows, and OpenClaw is a published first-party AgentSEO plugin.

Browse all integrations

No-code automation

n8n

Run AgentSEO inside n8n with HTTP Request nodes so SEO checks, content-gap jobs, and rank alerts can flow into broader automations.

HTTP workflowOpen guide

No-code automation

Make

Connect AgentSEO to Make with HTTP modules so search-intelligence jobs can trigger CRM updates, alerts, reports, and client handoffs.

HTTP workflowOpen guide

MCP client

Claude Code

Use AgentSEO inside Claude Code through MCP so the model can call SEO tools directly instead of scraping pages and guessing at next steps.

Hosted MCP or local MCPOpen guide

MCP client

Claude Desktop

Use AgentSEO in Claude Desktop through the local MCP package so Claude can run SEO workflows without manual copying between tools.

Local MCP packageOpen guide

Agent runtime

OpenClaw

Install the first-party AgentSEO OpenClaw plugin to give OpenClaw agents direct access to SEO and GEO workflows with allowlisted tools.

First-party AgentSEO pluginOpen guide

Builder-marketer path

Pair Claude Code with AgentSEO when one person spans strategy and implementation.

This path works best when you need SERP triage, workflow prototypes, small internal tools, and repeatable review loops instead of one-off prompting.

See Claude Code path

Start safe

How vibe marketers can use Claude Code for SEO workflows

Build the loop

From prompt to monitored workflow

Operationalize

Builder-marketer workflow with Claude Code, AgentSEO, and docs

Why this exists

SEO data wasn't built for the agent era.

Scraping gets you text. Dashboards get you charts. Agents need SERP-aware intelligence in a deterministic shape.

Bloated responses

Traditional SEO APIs ship dashboard payloads. Agents need strict schemas, not report dumps.

Wasted spend

Parsing provider output burns tokens and engineering time. Every workaround becomes integration debt.

Agents deserve better

SEO infrastructure should serve autonomous workflows. Deterministic outputs with sources beat brittle scraping and guesswork.

Try Live Demo See endpoints

Try it

Run a real check with copy and paste

Paste a key, run one request, and see the exact output (JSON + markdown_summary).

Get API Key

quicktest.sh

curl

$ curl /api/v1/local-visibility/track

export AGENTSEO_API_KEY="sk_live_your_key"

curl -sS -X POST "https://www.agentseo.dev/api/v1/local-visibility/track?sync=true" \
  -H "x-api-key: $AGENTSEO_API_KEY" \
  -H "content-type: application/json" \
  -d '{"domain":"example.com","keywords":["coffee near me"],"locations":["Austin, TX"]}'

Input → Output

Paste a key, run one request, and see the exact response shape.

Run with API key

API key

Domain

Keyword(s)

comma-separated

Location(s)

comma-separated

example response

# Local Visibility (Austin, TX)

**Domain**: northbeamcoffee.com  
**Overall visibility score**: 86.67

## Snapshots
- coffee near me (Austin, TX): rank #2 (95)
- coffee shop austin (Austin, TX): rank #4 (85)
- best coffee austin (Austin, TX): rank #6 (80)

## Next actions
- Improve proximity + relevance signals for the lowest-scoring keyword.
- Earn 3–5 fresh citations from local publications / directories.
- Publish a location page that matches high-intent phrasing.

What you get

Ship SEO intelligence. Not scrapers.

Give agents deterministic SEO intelligence they can act on, with response shapes built for tool use and automation.

Agents that reason over real SERP data

Your LLM gets structured JSON: local rankings, content gaps, SERP analysis, and competitor context. No parsing. No guessing.

Token-optimized outputs per call

Compact schemas keep your context window focused on reasoning, not cleaning up dashboard-shaped payloads.

One MCP server. Broad endpoint surface.

Use REST, MCP, or SDKs across 14 workflows, from local audit to content decay detection and AI overview extraction.

You build the agent. We handle SERP data.

Stable response shapes, typed SDKs, and async jobs with polling. Your engineering time goes to product, not plumbing.

How to use it

Your agent handles SEO. You ship product.

Already ran the live demo? Pick a workflow to ship next. Each check returns structured JSON your agent can reason over.

Submit URL + keyword

POST /api/v1/content/gap?sync=true

Fetch SERP competitors

sync=true (inline result)

Extract outlines

scrape_depth: h1/h2/h3

Return structured output

JSON + markdown_summary

response.json

{
  "keyword": "<string>",
  "user_url": "<url>",
  "missing_topics": ["<string>"],
  "competitor_outlines": [
    { "url": "<url>", "headers": ["<string>"] }
  ],
  "markdown_summary": "<markdown>"
}

Each response includes a markdown_summary you can pipe to Slack, email, or a task queue.

Get Free API Key →Read the OpenAPI spec →

Pricing

Start free. Upgrade when it sticks.

No seat licensing. Get one-time free credits to validate quickly, then upgrade for monthly usage.

View full pricing

Hobby

Try the live API with free credits (no card)

$0/month

100 live credits (one-time)
Mock mode + playground
No credit card required

Typical volume: ~100 searches or ~10 content gaps

Start Free

Starter

Ship real integration

Pro

Single product in active production

$49/month

10,000 credits/month
Higher throughput + priority queue
Upgrade path to Agency

Typical volume: ~10,000 searches or ~1,000 content gaps

Start Pro

Agency

Multi-client operations and reporting

$149/month

35,000 credits/month
Reserved throughput + multi-client workflows
Dedicated support for scale

Typical volume: ~35,000 searches or ~3,500 content gaps

Contact Sales

Who's building with this

Built for teams turning SEO work into repeatable systems

The common pattern is simple: get grounded search state, route it into a workflow, and keep the output usable by both operators and agents.

Builder

Agent-native SEO workflows

Ship tool calls that return structured SERP outputs your agent can reason over. No scraping glue code.

Uses: REST/MCP + strict JSON

Consultant

Gap checks you can send to a writer

Run a content gap check and share the markdown_summary as a brief. Fast review, clear next steps.

Uses: /api/v1/content/gap + markdown_summary

Growth

Weekly local visibility routines

Track visibility across keywords and locations and turn snapshots into a repeatable weekly check-in.

Uses: /api/v1/local-visibility/track + snapshots

Usage-based pricing with hard caps

Fast enough for weekly operating loops

Human review before higher-risk automation

FAQ

Answers before you ship

Short, practical answers. No sales fluff. If you still have questions, the OpenAPI spec is the source of truth.

Read OpenAPI

Quick facts

Safe retries with idempotency-key support
Predictable error payloads with request IDs

Scraping gets you page text. AgentSEO gives SERP-aware outputs your workflows can act on: rankings, competitor context, and structured summaries with deterministic shapes.

Get started

Make your first request with one curl

Copy a curl, paste your API key, and see the exact response shape (JSON + markdown_summary). Then install the SDK or MCP server when you’re ready.

Try Live Demo

# 1) Set your API key
export AGENTSEO_API_KEY="sk_live_your_key"

# 2) Run a first request (sync)
curl -sS -X POST "https://www.agentseo.dev/api/v1/local-visibility/track?sync=true" \
  -H "x-api-key: $AGENTSEO_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "domain": "northbeamcoffee.com",
    "keywords": ["coffee near me"],
    "locations": ["Austin, TX"]
  }'

SEO API for AI Agents and Developers

SEO API your agentscan actually use

Pick the job first, then take the shortest path

Make one successful request

Pick the workflow bundle

Find the client-facing workflows

Validate fit with one real workflow, not a generic demo

Compare API shape before you compare endpoint count

See what a safe SEO agent loop actually looks like

Build the AI visibility loop before you build the dashboard

Pick the runtime, then copy the shortest proven path

n8n

Make

Claude Code

Claude Desktop

OpenClaw

Pair Claude Code with AgentSEO when one person spans strategy and implementation.

SEO data wasn't built for the agent era.

Bloated responses

Wasted spend

Agents deserve better

Run a real check with copy and paste

Ship SEO intelligence. Not scrapers.

Agents that reason over real SERP data

Token-optimized outputs per call

One MCP server. Broad endpoint surface.

You build the agent. We handle SERP data.

Your agent handles SEO. You ship product.

Find what is missing

Track rank by city

Catch decay early

Start free. Upgrade when it sticks.

Hobby

Starter

Pro

Agency

Built for teams turning SEO work into repeatable systems

Answers before you ship

Make your first request with one curl

SEO API your agentscan actually use

SEO data wasn't built for the agent era.

Bloated responses

Wasted spend

Agents deserve better

Run a real check with copy and paste

Ship SEO intelligence. Not scrapers.

Agents that reason over real SERP data

Token-optimized outputs per call

One MCP server. Broad endpoint surface.

You build the agent. We handle SERP data.

Your agent handles SEO. You ship product.

Find what is missing

Track rank by city

Catch decay early

Make your first request with one curl