Documentation Center

This is the developer and integration documentation for MCP, API authentication, parameters, cost estimates, and responses. If you are connecting SEO Agent for the first time, start with the beginner setup guide.

Getting Started Guide

SEO Agent provides capabilities in two ways: MCP lets Codex and other AI tools call SEO capabilities from natural-language goals, while the API supports scripts, client reports, and internal systems. Both use the same SEO Agent API Key and public billing rules.

Hands-on Tutorial

Step 1: Register an Account

Visit SEO Agent Homepage, click the "Sign Up / Log In" button in the top right corner, fill in your email, password, and confirm password to complete registration. The registration process is simple and fast, taking less than 1 minute.

Step 2: Add Balance

After logging in, open the Pricing page, select a top-up amount, and complete payment. SEO Agent uses a pay-as-you-go model with a minimum top-up of $30. Your balance remains valid and does not expire.

Note: We recommend starting with a small top-up to try it out. Once you confirm it meets your needs, add more based on your usage. In principle, recharged amounts are non-refundable. Please proceed with caution.

Step 3: Start Using

To start SEO tasks in everyday language inside Codex, configure MCP. To integrate scripts, client reports, or internal systems, use the API. See the How to Use guide for the full beginner setup.

API Overview

SEO Agent provides a complete REST API for scripts, client reports, and internal systems. MCP and API share the same SEO Agent API Key: MCP is designed for automatic calls from AI tools, while the API is designed for stable calls from systems and scripts. Both return standardized data and billing details; Codex or your own AI workflow can turn those results into a complete consultant report.

Base URL:https://www.seoagent.vip/api/dev/

Default market: Except for site audit, if the request does not include location and language, the system defaults to United States / English for the query. When querying other markets, please pass it explicitly.

Recommended Call Flow

  1. On the profile page, create an API Key and select the permitted API scope.
  2. Decide how to integrate: MCP for AI tools, API for scripts, reports, and internal systems.
  3. Before making a formal query, first call /api/dev/estimate or pass in dryRun: true to estimate the cost.
  4. After confirming the parameters, market, and balance, execute the formal query, and from the response's data and billing read the results and billing information.

Supported API Endpoints

  • MCP Integration - Let Codex and other AI tools call SEO Agent based on your goal, without the need to manually remember endpoint paths.
  • API Integration - Integrate standardized SEO data into scripts, client reports, and internal systems
  • GEO / AI Search Visibility - Query Google AI Overview and ChatGPT for current mentions, top sources, historical trends, new/lost, multi-target comparison, and brand-category landscape
  • GEO Deep Audit - Check page citability, AI crawlers, llms.txt, Schema, and brand platform recommendations, with optional PDF report output
  • Keyword Research - Query search volume, competition, search intent, and related keywords for seed keywords
  • In-depth Keyword Research - Complete KD, CPC, trends, question keywords, commercial value, and keyword grouping
  • Domain-specific keyword opportunities - Based on the target domain's current rankings, decide whether to optimize existing pages or create new pages
  • Domain Keywords - Query the current organic ranking keywords for a specified domain
  • Competitor Keyword Gap - Compare keyword gaps between your website and competitor domains
  • Competitor keyword strategy analysis - Enter seed keywords to compile high-search-volume keywords from top-ranking peers and competitors
  • Keyword ranking lookup - Query the organic ranking of a specified domain for a given keyword
  • On-site audit - Use SEO Agent's built-in crawler to audit technical SEO and on-site SEO
  • GSC Search Performance and Diagnostics - Read clicks, impressions, CTR, rankings, content decay, keyword cannibalization, and action plans for authorized sites
  • GA4 Traffic Quality Diagnostics - Read sessions, channels, landing pages, engagement rate, bounce rate, and key events for authorized media resources, and output landing page opportunities, channel diagnostics, and action plans
  • Account Balance - Query the account balance and available quota for the current API Key
  • Spending History - Query historical API call consumption records

Authentication Methods

All API requests must include an API Key in the request header for authentication. You can create and manage API Keys on your personal settings page.

POST /api/dev/{endpoint}

Request Header

ParameterTypeRequiredDescription
AuthorizationstringYesBearer {your_api_key}
Content-TypestringYesapplication/json

Example Code

curl -X POST https://www.seoagent.vip/api/dev/keyword-research \
  -H "Authorization: Bearer your_api_key_here" \
  -H "Content-Type: application/json" \
  -d '{
    "keyword": "running shoes",
    "location": "United States",
    "language": "English"
  }'

Cost Estimate API

Estimate the maximum cost before making a formal query. No query is executed, and no charge is incurred. Supports cost estimation for all query-type APIs.

POST /api/dev/estimate

Request Parameters

ParameterTypeRequiredDescription
endpointstringYesThe name of the API to estimate, such as keyword-research, site-audit, etc.
paramsobjectYesThe request parameter object for the corresponding API

Note: You can also pass in when calling any query API dryRun: true the parameter to get the estimated cost without actually executing the query.

Example: Estimate to Formal Call

# 1. Estimate the cost first
curl -X POST https://www.seoagent.vip/api/dev/estimate \
  -H "Authorization: Bearer your_api_key_here" \
  -H "Content-Type: application/json" \
  -d '{
    "endpoint": "domain-keyword-opportunities",
    "params": {
      "domain": "example.com",
      "keyword": "running shoes",
      "location": "United States",
      "language": "English",
      "limit": 10
    }
  }'

# 2. Make the formal query after confirmation
curl -X POST https://www.seoagent.vip/api/dev/domain-keyword-opportunities \
  -H "Authorization: Bearer your_api_key_here" \
  -H "Content-Type: application/json" \
  -d '{
    "domain": "example.com",
    "keyword": "running shoes",
    "location": "United States",
    "language": "English",
    "limit": 10
  }'

Keyword Research API

Query the search volume, competition, search intent, and related keywords for a seed keyword, helping you determine which keywords are worth prioritizing.

POST /api/dev/keyword-research

Request Parameters

ParameterTypeRequiredDescription
keywordstringYesTarget keyword (choose either this or keywords)
keywordsarrayNoArray of batch keywords, deduplicated first
locationstringNoTarget market, default is United States
languagestringNoTarget language, default is English
limitnumberNoResult count limit
dedupeModestringNoSimilar keyword deduplication mode: "all" or "smart"

Deep Keyword Research API

On top of basic keyword research, it adds keyword difficulty, CPC, trends, question keywords, commercial value, suggested page types, and keyword grouping. Suitable for content planning and page matrix creation.

POST /api/dev/keyword-intelligence

Request Parameters

ParameterTypeRequiredDescription
keywordstringYesSeed keyword or business topic
locationstringNoTarget market, default is United States
languagestringNoTarget language, default is English
limitnumberNoNumber of keywords to return

Main response fields

  • keywords: Keyword details, including search volume, CPC, competition, KD, trend score, question keywords, commercial score, suggested page type, and priority.
  • groups: Keyword grouping, including group name, main keyword, total search volume, average difficulty, number of commercial keywords, number of question keywords, and recommended page type.
  • history: Keyword historical trend summary.

Domain Personalized Keyword Opportunity API

Enter a target domain and seed keyword. SEO Agent combines keyword expansion data with the target domain's existing organic rankings to determine which keywords the website should prioritize, whether matching pages already exist, and whether to recommend optimizing existing pages or creating new pages.

POST /api/dev/domain-keyword-opportunities

Request Parameters

ParameterTypeRequiredDescription
domainstringYesTarget website domain
keywordstringYesSeed keyword or business topic
locationstringNoTarget market, default is United States
languagestringNoTarget language, default is English
limitnumberNoNumber of opportunity keywords to return
domainKeywordLimitnumberNoNumber of the domain's ranking keyword samples used for matching, default 50, maximum 100

Main response fields

  • keywords: Personalized opportunity keywords, including opportunity score, current ranking, matching page, difficulty fit score, recommended action, and P0/P1/P2 priority.
  • groups: Keyword groups merged by topic, helping determine how many pages to create.
  • domainMetrics: Ranking capability metrics from the target domain's current sample.

Domain Keyword API

Query the current organic ranking keywords of a specified domain to understand the website's keyword coverage.

POST /api/dev/domain-keywords

Request Parameters

ParameterTypeRequiredDescription
domainstringYesTarget domain
locationstringNoTarget market, default is United States
languagestringNoTarget language, default is English
limitnumberNoResult count limit

Competitor Keyword Gap API

Compare the keyword gap between your website and a competitor domain to find keyword opportunities where the competitor ranks but you don't.

POST /api/dev/competitor-gap

Request Parameters

ParameterTypeRequiredDescription
domainsarrayYesDomain array (two domains: the first is yours, the second is a competitor)
domainstringNoYour domain (used with competitor)
competitorstringNoCompetitor domain (used with domain)
locationstringNoTarget market, default is United States
languagestringNoTarget language, default is English

Competitor Keyword Strategy API

Enter a seed keyword to automatically compile up to 5 top-ranking peer websites and up to 10 high-search-volume ranking keywords per competitor.

POST /api/dev/competitor-keyword-strategy

Request Parameters

ParameterTypeRequiredDescription
keywordstringYesSeed keyword
locationstringNoTarget market, default is United States
languagestringNoTarget language, default is English
competitorLimitnumberNoMaximum number of competitors, default 5, standard plan limit 5
keywordsPerCompetitornumberNoMaximum number of keywords per competitor, default 10, standard plan limit 10

Ranking Query API

Query the organic ranking position of a specified domain for a given keyword.

POST /api/dev/serp-rank-query

Request Parameters

ParameterTypeRequiredDescription
keywordstringYesTarget keyword
domainstringYesTarget domain
locationstringNoTarget market, default is United States
languagestringNoTarget language, default is English
limitnumberNoMaximum number of rankings to query, default 30, maximum 100.

GEO Visibility API

Use the new LLM Mentions to query current mentions, target metrics, high-frequency source domains and pages for Google AI Overview or ChatGPT. Can be via platform Select google or chat_gpt, via lite: true use lightweight query.

POST /api/dev/geo-visibility

Request Parameters

ParameterTypeRequiredDescription
topicstringYesAnalyze topic, brand, or domain (choose one of three, mutually exclusive with keyword/domain)
keywordstringNoKeywords
domainstringNoDomain
locationstringNoTarget market, default is United States
languagestringNoTarget language, default is English
platformstringNogoogle or chat_gpt, default google
litebooleanNoWhether to use Lite endpoint, default false

GEO Trends, Comparison, and Brand Landscape API

The following endpoints all use the new LLM Mentions and no longer call the old endpoints. The default platform is google; specify chat_gpt only supports United States / English.

POST/api/dev/geo-visibility-history

Historical monthly trends

Required analysis object: topickeyword or domain Choose one of three. Optional platformdateFromdateTolocationlanguage.

{
  "domain": "yourdomain.com",
  "platform": "google",
  "dateFrom": "2026-01-01",
  "dateTo": "2026-07-01",
  "location": "United States",
  "language": "English"
}
POST/api/dev/geo-visibility-delta

Change amount

Used to view the magnitude of changes in mentions and AI search volume. groupRange Optional weekmonth or year.

POST/api/dev/geo-visibility-new-lost

New and lost mentions

Parameters are the same as the change amount endpoint, and it returns the newly added and lost mentions for each period along with the corresponding AI search volume.

POST/api/dev/geo-visibility-comparison

Multi-target visibility comparison

targets Must provide 2 to 10 brands, domains, or topics. All objects use the same platform, market, and language, and the relative share in this dataset is returned.

{
  "targets": ["yourdomain.com", "competitor-a.com", "competitor-b.com"],
  "platform": "google",
  "location": "United States",
  "language": "English"
}
POST/api/dev/geo-brand-landscape

Brand and category landscape

Query the most frequently appearing brands and brand categories under a topic. Can pass lite: true Use the Lite endpoint to reduce query costs when details are not needed.

AI mentions, citations, AI search volume, and relative share are not equivalent to actual website visits or conversions. Real traffic and business results still need to be verified with GA4, server logs, and key events.

GEO Deep Audit API

Audit the GEO foundation of a single website or page, including page citability, AI crawler access status, llms.txt structure, Schema suggestions, brand platform suggestions, and an overall score. llms.txt It is a supplementary explanatory document and does not mean that AI search will necessarily cite it; the report will separate detected facts from optimization suggestions.

POST /api/dev/geo-deep-audit

Request Parameters

ParameterTypeRequiredDescription
urlstringYesTarget page URL or domain; the domain will be auto-completed to https://
domainstringNoTarget domain, used for report identification
brandNamestringNoBrand name, used for brand platform suggestions
maxPagesnumberNoThe page limit for generating llms.txt suggestions; default 20, maximum 50
includePdfReportbooleanNoWhether to return a base64 PDF report in the response; default false

Returned data

  • citability: Page paragraph citability score and low-scoring paragraph clues.
  • crawlers: Access status of major AI crawlers in robots.txt.
  • llmsSections: Whether llms.txt exists, and the section and link structure.
  • schemaTypes: Structured data of the current page and suggested supplementary types.
  • pdfReport: Only when includePdfReport: true returned, can be saved as a PDF file.

Site Audit API

Use SEO Agent's built-in crawler, first read robots.txt and Sitemap to estimate the number of pages, then audit website pages, titles, descriptions, H1, canonical, internal/external links, image alt, response speed, and indexing issues. For 100 pages or fewer, it executes synchronously after fee confirmation; for more than 100 pages, it automatically creates a background asynchronous task after fee confirmation.

POST /api/dev/site-audit

Request Parameters

ParameterTypeRequiredDescription
urlstringYesTarget website URL or domain
domainstringNoDomain
limitnumberNoMaximum number of pages to audit; when a Sitemap exists, defaults to the number of discovered pages, with a safety limit of 5000
directorystringNoOnly audit the specified directory, e.g. /products/; suitable for use when the balance is insufficient or you want to audit by directory.
confirmAuditbooleanNoFor up to 100 pages, set to true after the customer confirms the maximum fee returned by the pre-check.
confirmFullAuditbooleanNoFor over 100 pages, set to true after the customer confirms the maximum fee and estimated time for the asynchronous audit.
maxDepthnumberNoCrawl depth, default 2, max 5.
respectRobotsbooleanNoWhether to respect robots.txt, default true.
includeSubdomainsbooleanNoWhether to include subdomains, default false.
renderJavascriptbooleanNoWhether to enable JS rendering, default false.
renderWaitMsnumberNoJS rendering wait time (milliseconds), default 0, max 10000.
targetKeywordsarrayNoTarget keyword array, used for content relevance analysis.
duplicateSimilarityThresholdnumberNoDuplicate content similarity threshold, default 0.86, range 0.65-0.98.

Site Audit (Async) API

Websites with more than 100 pages use asynchronous site audit: SEO Agent continuously crawls Sitemap pages in the background and immediately returns a task ID. The returned result explains the estimated time, maximum fee, and how to query; it supports receiving completion notifications via callbackUrl, or obtaining real-time progress and the final report through the query API.

POST /api/dev/site-audit-async

Request Parameters

ParameterTypeRequiredDescription
urlstringYesTarget website URL or domain
callbackUrlstringNoCallback notification address when the task completes.
confirmFullAuditbooleanYesSet to true after the customer confirms the maximum fee and estimated time of the pre-check.
Other parameters are the same as the Site Audit API.

Callback security: callbackUrl must be a public HTTP/HTTPS address and cannot point to localhost or a private network. SEO Agent sends X-SEOAgent-Callback-Timestamp and X-SEOAgent-Callback-Signature. The signature is HMAC-SHA256; its key is the SHA-256 digest of the current API key, and the signed value is “timestamp.raw request body”. Reject stale timestamps and verify the signature with a constant-time comparison. In production, use DEVELOPER_CALLBACK_ALLOWED_HOSTS for a comma-separated callback host allowlist.

The callback body may contain website analysis results. Only use an endpoint you control; do not send results to an unfamiliar third-party URL.

GET /api/dev/site-audit-async/{taskId}

Path parameters

ParameterTypeRequiredDescription
taskIdstringYesAsync task ID

Task status

  • pending - Pending execution
  • processing - Processing
  • completed - Completed
  • failed - Failed

progress It returns the number of discovered pages, target pages, processed pages, successful pages, failed pages, completion percentage, and estimated completion time. When the status is completed after, data the field contains the final report; the report clearly states how many pages were discovered, how many succeeded, how many failed, and whether it is complete.

GSC Search Performance and Diagnostics API

Read GSC search performance data for the user's authorized properties, used to integrate real clicks, impressions, CTR, average position, content decay, keyword cannibalization, and prioritized action plans into Codex, scripts, or internal reports. Before using, you must first complete the GSC connection on the profile page and check the corresponding GSC permissions when creating an API Key.

POST /api/dev/gsc-sites

Purpose

Return the list of GSC properties authorized for the current account and the default property.

POST /api/dev/gsc-search-analytics

Request Parameters

ParameterTypeRequiredDescription
siteUrlstringNoGSC property URL. If not provided, the default property selected on the profile page is used.
startDatestringNoStart date, format YYYY-MM-DD, defaults to the last 28 days.
endDatestringNoEnd date, format YYYY-MM-DD, defaults to yesterday.
dimensionsarrayNoDimensions, e.g., query, page, country, device, date, searchAppearance.
rowLimitnumberNoNumber of rows to return, default 1000, maximum 25000.
searchTypestringNoweb, image, video, news, discover, or googleNews. Default is web.

Example

curl -X POST https://www.seoagent.vip/api/dev/gsc-search-analytics \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "dimensions": ["query", "page"],
    "startDate": "2026-06-01",
    "endDate": "2026-06-28",
    "rowLimit": 1000
  }'
POST /api/dev/gsc-keyword-opportunities

Purpose

Based on real GSC search performance, identify keyword opportunities such as high impressions with low CTR, positions 8-20, etc., and return. opportunities.

POST /api/dev/gsc-page-opportunities

Purpose

Aggregate GSC query performance by page, return page opportunities that need priority optimization of titles, content coverage, and internal links, return. pages.

POST /api/dev/gsc-performance-summary

Purpose

Compare clicks, impressions, CTR, and average position between the current period and the previous period of equal length, return. summarytimeseriestopQueriestopPagesqueryMovers and pageMovers.

POST /api/dev/gsc-content-decay

Purpose

Identify pages and queries with significant declines in clicks or rankings, return. decayedPages and decayedQueries, suitable for content refresh and index recheck.

POST /api/dev/gsc-cannibalization

Purpose

Discover cases where the same query is handled by multiple pages, return. clusters, used to determine whether to designate a main page, merge content, adjust canonical, or differentiate content.

POST /api/dev/gsc-action-plan

Purpose

Combine GSC keyword opportunities, page opportunities, content decay, and keyword cannibalization, return. actions, each action includes priority, evidence, recommended actions, and review metrics.

GA4 Traffic Quality Diagnostic API

Read GA4 data for the user's authorized properties, used to integrate sessions, users, channels, landing pages, engagement rate, bounce rate, key events, and conversion attribution diagnostics into Codex, scripts, or internal reports. Before using, you must first complete the GA4 connection on the profile page and check the corresponding GA4 permissions when creating an API Key.

POST /api/dev/ga4-properties

Purpose

Return the list of GA4 properties authorized for the current account and the default property.

POST /api/dev/ga4-report

Request Parameters

ParameterTypeRequiredDescription
propertyIdstringNoGA4 property ID. If not provided, the default property selected on the profile page is used.
startDatestringNoStart date, format YYYY-MM-DD, defaults to the last 28 days.
endDatestringNoEnd date, format YYYY-MM-DD, defaults to yesterday.
dimensionsarrayNoDimensions, e.g., landingPagePlusQueryString, sessionDefaultChannelGroup, country, deviceCategory, date.
metricsarrayNoMetrics, e.g., activeUsers, sessions, screenPageViews, engagementRate, bounceRate, keyEvents.
limitnumberNoNumber of rows to return, default 1000, maximum 100000.

Example

curl -X POST https://www.seoagent.vip/api/dev/ga4-report \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "dimensions": ["landingPagePlusQueryString"],
    "metrics": ["sessions", "engagementRate", "bounceRate", "keyEvents"],
    "startDate": "2026-06-01",
    "endDate": "2026-06-28",
    "limit": 100
  }'
POST /api/dev/ga4-traffic-summary

Purpose

Summarize sessions, users, views, key events, channel, device, and country performance, return. summarytimeserieschannelsdevicescountries and insights.

POST /api/dev/ga4-landing-page-opportunities

Purpose

Identify landing pages with visits but poor engagement, high bounce rate, or insufficient key events, and return pages Each page includes priority, evidence, and optimization suggestions.

POST /api/dev/ga4-channel-insights

Purpose

Break down sessions, engagement, bounce, and key events by channel, and return channels This is used to determine which sources need a review of landing quality.

POST /api/dev/ga4-action-plan

Purpose

Combining GA4 traffic summary, landing page opportunities, and channel quality diagnosis, return actions Each action includes P0/P1/P2 priority, evidence, suggested action, and review metrics.

Account Balance API

Query the account balance, available credit, and plan information associated with the current API Key.

POST /api/dev/balance

Request Parameters

ParameterTypeRequiredDescription
No request parameters

Usage Records API

Query the historical usage records of API calls, including endpoint type, call time, charged amount, and call status.

POST /api/dev/usage-records

Request Parameters

ParameterTypeRequiredDescription
limitnumberNoNumber of records to return, default 30, maximum 200

What is an SEO workflow?

An SEO workflow breaks a natural-language task into multiple sequential steps. For example, when you say “create a content set around CRM software suitable for small businesses,” SEO Agent first looks up keywords and search results, then plans core pages, supporting articles, and internal links, rather than directly writing an article from scratch.

You don't need to remember professional commands. Just describe the website, product or topic, and target market, and the system will choose an appropriate workflow; if necessary information is missing, it will ask you first. After query, analysis, and plan generation are complete, the workflow pauses at the confirmation stage. Only when you approve the results and explicitly give approval will the system continue to modify files or publish to an already connected website.

Billing notes: The workflow is billed by the stages actually completed. Before each stage runs, its purpose and estimated cost are shown; when the workflow is paused, waiting for additional information, or waiting for confirmation, no repeated charges are made.

A complete usage example

  1. In the Codex conversation, say: “Help me plan a content cluster around CRM software suitable for small businesses.”
  2. SEO Agent looks up keywords and search results, and returns core pages, supporting articles, search intent, and internal link suggestions.
  3. You review the results. If adjustments are needed, you can continue: “Change the focus to teams of fewer than 10 people in the United States.”
  4. After approval, say: “Confirm this plan and publish the first article to my connected WordPress draft folder.”
  5. After publishing, the system returns the page link; if monitoring has been set up, it will also re-check on schedule and send notifications.

Five common workflow templates

TemplateWhen to useWhat the system will accomplish
New site SEO launchA new website has just launched, and you don't know what to do first.Check the technical foundation, research keywords, and plan the first batch of pages.
Competitor catch-upWant to know what traffic opportunities competitors have.Compare keyword gaps, identify pages you can capture, and prioritize them.
Content cluster planningWant to consistently create content around a single product.Plan core pages, supporting articles, search intent, and internal links.
Technical audit fixesPossible issues with site indexing, speed, or page structure.Audit issues, generate a P0/P1/P2 list, then fix and verify after approval.
GEO visibility improvementWant ChatGPT and other AI search engines to mention the brand more easily.Establish a visibility baseline, plan page citations, and improve brand information.

Also possible after logging in. GET /api/workflow-templates Read template, approve. POST /api/workflows Create a workflow, and use POST /api/workflows/{id}/approve Approve to continue execution. Usually there is no need to manually call these APIs; Codex automatically creates and advances workflows based on the conversation.

Approval, publishing, and scheduled monitoring

Connect WordPress or CMS

  1. Log in to SEO Agent and go to “Profile”.
  2. Find “Publishing Connection” and select WordPress or Generic CMS.
  3. Enter the website address and the dedicated access credentials provided by that website, then click “Test Connection”.
  4. After the test succeeds, return to the Codex conversation to create a content workflow. The system will still wait for your confirmation before publishing.

It is recommended to create a separate publishing account with restricted permissions for SEO Agent and prioritize publishing as drafts. After deleting the connection, SEO Agent will no longer retain the access entry for that connection.

Scheduled monitoring and notifications

After completing a workflow, you can set it as a scheduled task. For example, review technical issues weekly, and check keyword and GEO visibility changes monthly. The system will copy the original workflow’s goals and inputs to re-execute, and write the status of completed, failed, or awaiting manual confirmation to the notification center; after configuring an email service, it can also send email reminders.

Relevant APIs include POST /api/workflow-schedulesGET /api/workflow-schedules and GET /api/notifications Scheduled tasks will not skip approval: when page modifications or formal publishing are involved, manual confirmation is still required according to the workflow settings.

MCP Overview

MCP (Model Context Protocol) is a protocol that allows AI tools to directly call external capabilities. With MCP integration, MCP-capable AI tools (such as Codex) can automatically understand your goals, proactively call SEO Agent's standardized data capabilities, and complete continuous workflows including querying, organizing, and preparing analysis drafts. For standalone website sellers, SEO agencies, and foreign trade businesses, its value is to reduce the time spent manually entering multiple tools to check data, copy, and organize tables.

MCP Advantages

  • Automated workflows - AI tools automatically call capabilities based on goals, with no need to manually switch tools
  • Better suited to SEO agencies for batch delivery - Can organize multiple clients' keyword, GEO, competitor, and authorized data into drafts
  • Better suited to standalone website growth - Can continuously find page opportunities around product keywords, target countries, and competitors
  • Natural language interaction - Describe tasks in natural language, and the AI executes automatically after understanding

Supported MCP tools

  • seoagent_keyword_research: Lightweight keyword research, suitable for quickly checking search volume, competition, intent, and related keywords.
  • seoagent_keyword_intelligence: In-depth keyword research, returns KD, CPC, trends, question keywords, commercial value, and keyword grouping.
  • seoagent_domain_keyword_opportunities: Domain-specific keyword opportunities, combined with the domain's existing rankings to determine whether to optimize existing pages or create new pages.
  • seoagent_domain_keywords: Query the existing organic ranking keywords for a specified domain.
  • seoagent_competitor_gap: Compare the keyword gap between our domain and competitor domains.
  • seoagent_competitor_keyword_strategy: Based on seed keywords, compile high-search-volume keywords from top-ranked peers and competitors.
  • seoagent_serp_rank_query: Query the ranking position of a specified domain for a specified keyword.
  • seoagent_geo_visibility: Query the GEO / AI search visibility of a brand, topic, or domain.
  • seoagent_geo_visibility_history: Query historical AI visibility trends.
  • seoagent_geo_visibility_delta: Query the change in AI visibility.
  • seoagent_geo_visibility_new_lost: Query new and lost AI mentions.
  • seoagent_geo_visibility_comparison: Compare AI visibility for up to 10 targets.
  • seoagent_geo_brand_landscape: Query high-frequency AI brands and brand category landscape.
  • seoagent_geo_deep_audit: Perform a deep GEO audit, returning page citability, AI crawlers, llms.txt, Schema, and an optional PDF report.
  • seoagent_site_audit: First read the Sitemap to return estimated page count, maximum cost, and estimated time; within 100 pages, run a synchronous audit after confirmation; over 100 pages, switch to background asynchronous batch audit after confirmation.
  • seoagent_get_site_audit_task: Use the task ID to query async audit progress; after task completion, read the final report containing a completeness statement.
  • seoagent_gsc_sites: Query the authorized GSC site list and default site.
  • seoagent_gsc_search_analytics: Query GSC search performance, including clicks, impressions, CTR, and average position.
  • seoagent_gsc_keyword_opportunities: Identify high-impression low-CTR and Quick Win keyword opportunities.
  • seoagent_gsc_page_opportunities: Identify page opportunities that require priority optimization.
  • seoagent_gsc_performance_summary: Compare GSC performance between the current period and the previous period, outputting sources of growth and decline.
  • seoagent_gsc_content_decay: Discover pages and queries with significantly declining clicks or rankings.
  • seoagent_gsc_cannibalization: Identify keyword cannibalization risks where multiple pages target the same query.
  • seoagent_gsc_action_plan: Generate a P0/P1/P2 priority action list based on comprehensive GSC diagnostic results.
  • seoagent_ga4_properties: Query the authorized GA4 property list and default property.
  • seoagent_ga4_report: Query GA4 report data, including sessions, users, views, engagement rate, bounce rate, and key events.
  • seoagent_ga4_traffic_summary: Summarize GA4 traffic quality, channel, device, and country performance.
  • seoagent_ga4_landing_page_opportunities: Identify landing page opportunities with visits but weak conversion or follow-through.
  • seoagent_ga4_channel_insights: Diagnose channel traffic quality and key event performance.
  • seoagent_ga4_action_plan: Synthesize GA4 diagnostic results to generate a P0/P1/P2 priority action list.
  • seoagent_search_consultant_knowledge: On-demand retrieval of SEO consultant methodology, scenario judgment, and official practice summaries; can be automatically used in strategy, diagnostic, and reporting tasks. MCP will require AI to present results in natural, easy-to-understand client language rather than directly displaying tool data.
  • seoagent_get_consultant_knowledge_card: Read a retrieved knowledge card when finer decision boundaries, execution workflows, or acceptance criteria are needed.
  • seoagent_estimate: Estimate costs and cache hit status before making a formal query.

MCP Configuration Guide

Prerequisites

  1. Have registered an SEO Agent account and completed the initial top-up.
  2. Created an API Key on the personal settings page.
  3. Use an AI tool that supports the MCP protocol (such as Codex).

Recommended: Complete configuration in a Codex conversation.

Open any trusted Codex conversation and the following YOUR_SEO_AGENT_API_KEY After replacing with your own SEO Agent API Key, send it directly. Codex will write the remote MCP configuration to your local machine's ~/.codex/config.toml, retaining the existing MCP configurations.

Please add the SEO Agent remote MCP to my local Codex.

Only modify ~/.codex/config.toml, and do not delete or overwrite existing configurations;
Append the following configuration to the end of the file, and check the TOML syntax after writing:

[mcp_servers.seoagent]
enabled = true
url = "https://www.seoagent.vip/mcp"

[mcp_servers.seoagent.http_headers]
Authorization = "Bearer YOUR_SEO_AGENT_API_KEY"

Do not echo my API Key in your reply. After completion, tell me whether I need to start a new conversation or restart Codex for it to load.

Security note: Only enter your API Key in your own local Codex conversation. Do not send the Key to group chats, screenshots, or paste it on public pages. For each Codex, script, or client project, we recommend using a separate Key, and set permissions, quota, and expiration.

If you prefer to edit the configuration file yourself, you can append the following content to ~/.codex/config.toml. This is the remote MCP format currently used by Codex, not the old mcpServers JSON or local Node startup script format.

[mcp_servers.seoagent]
enabled = true
url = "https://www.seoagent.vip/mcp"

[mcp_servers.seoagent.http_headers]
Authorization = "Bearer YOUR_SEO_AGENT_API_KEY"

Codex Integration Guide

After completing the configuration in a Codex conversation, you can use natural language to have Codex call SEO Agent to complete tasks such as standalone website SEO, GEO, keyword research, and client report drafts.

Note: After registration bonuses, redemption vouchers, or standard top-ups are credited, API / MCP can be enabled; when making calls, you still need to ensure sufficient account balance and API Key quota.

Loading and Verification

  1. After letting Codex complete the configuration, start a new conversation; if the tools do not appear, restart Codex.
  2. In the new conversation, send: Check whether the SEO Agent MCP is connected, and only list the names of the available SEO tools.
  3. After confirming the connection, send real business tasks; for the first time, it is recommended to use a small-scope keyword or domain query to verify permissions, market, and balance.

Troubleshooting: If Codex reports that it cannot find seoagent then ask it to read ~/.codex/config.toml and confirm that these exist [mcp_servers.seoagent]url and Authorization three items, then start a new conversation. Do not paste the API Key into ordinary task text.

Using Codex

After configuration, you can directly ask questions in Codex, for example:

  • "Help me analyze the SEO opportunities for the standalone website example.com in the U.S. English market"
  • "Check the search volume and competition for the keyword running shoes, and determine whether it is suitable for a product page or a blog page"
  • "Deep-dive analysis of running shoes keyword groupings, commercial keywords, and question keywords"
  • "Analyze example.com's personalized keyword opportunities around running shoes in the U.S. English market"
  • "Compare the content gap between example.com and competitor.com"
  • "Perform a site audit for example.com"
  • "Using my connected GSC data, find keyword opportunities with high impressions, low clicks, and rankings 8-20 in the last 28 days"
  • "Using my connected GA4 data, find landing pages that had visits but weak key events in the last 28 days"

FAQ

Are there rate limits on API calls?

Yes. The default API call rate limit is 240 calls per minute. If you need a higher rate limit, please contact our sales team for a custom enterprise plan.

How can I estimate API call costs?

You can use /api/dev/estimate endpoint to estimate call costs, and you can also set it during the call. dryRun: true Parameters for getting estimated costs without actually executing.

Are API Keys secure?

API Keys are stored encrypted. The full key is only displayed once at creation and cannot be viewed later. You can set different API permissions (scope), consumption limits, and validity periods for each key to reduce risk.

Which programming language SDKs are supported?

Currently, we provide a REST API. You can call it using any programming language that supports HTTP requests. The official SDK is under development. Stay tuned!