Add Screpy as a custom MCP app, complete browser-based OAuth, and use project-scoped tools inside ChatGPT conversations.
SEO MCP for your AI assistant
Connect ChatGPT, Claude, Codex or Cursor to your Screpy projects. Investigate crawl issues, Search Console performance, saved rankings, AI answers and backlinks in one conversation, then decide what to work on next.
Hosted server. Sign in with Screpy. No local installation.
Setup guides for the AI tools you already use
- Screpy: Hello! What would you like to investigate about your website today?
- You: Check the latest completed crawl for example.com. What should I fix first?
- Screpy: Start with 3 pages missing titles, a broken internal link and the homepage’s slow loading result. I’ll keep the completed crawl and the latest homepage performance check separate. Audit priorities: Missing titles: 3. Broken links: 1. Homepage check: 1.
- You: Show me the pages with missing titles.
- Screpy: /collections/linen /products/linen-shirt /products/linen-trousers All three are missing titles in the latest completed crawl. Pages missing titles: /collections/linen: High priority. /products/linen-shirt: High priority. /products/linen-trousers: High priority.
- You: Which one matters most for organic traffic?
- Screpy: Search Console puts /collections/linen first: impressions, but a low click-through rate. I’d review its title before the two product pages. Search Console · collection: Impressions: 12,400. Clicks: 186. CTR: 1.5%.
- You: Suggest a better title for that collection.
- Screpy: Linen Clothing: Shirts & Trousers | Example A draft that describes the collection. Review it before publishing. Search result preview: example.com › collections › linen: Linen Clothing: Shirts & Trousers | Example. Explore linen shirts and trousers for everyday comfort.
- You: Where is the broken link?
- Screpy: The collection links to an old product URL. Update the destination to the current product page. Broken internal link: /collections/linen: /products/linen-shirt-old. 404 Update its destination: /products/linen-shirt. 200
- You: What does the latest homepage performance check show?
- Screpy: The latest checked homepage has a slow LCP and measurable blocking time. These are laboratory results, not field INP. Review the main image and scripts before choosing a fix. Latest homepage check · lab data: LCP: 3.8 s. TBT: 240 ms. CLS: 0.08.
- You: How are our tracked keywords doing?
- Screpy: “Linen clothing” slipped from position 8 to 12. That’s context, not proof that the missing title caused it. Tracked keyword · recent positions: Linen clothing: 8 → 12.
- You: Are AI assistants mentioning us?
- Screpy: The stored AI Visibility answers mention competitors more often. I can show the prompts and citations behind those mentions. Stored AI mentions: Example: 2. Stored answers Linen Studio: 5. Stored answers Nord Loom: 4. Stored answers Sources: 2. stylejournal.example · linenguide.example
- You: Turn that into a practical action plan.
- Screpy: 1. Add the collection title. 2. Repair the broken internal link. 3. Review the homepage’s loading result. 4. Measure again after your team makes the fixes. These are recommendations. Nothing has been changed. Your first three tasks: Collection title: Add and review the draft. Internal link: Update its destination. Homepage performance: Inspect image and scripts.
- You: What should I track after those fixes?
- Screpy: After your team makes the changes, explicitly request and confirm a new crawl to verify titles and links. Review the next homepage performance check, then compare Search Console periods and saved rankings as each source updates. After the fixes: New crawl: Verify titles and links. Search Console: Compare clicks and CTR. Performance & rankings: Compare refreshed results.
Connect AI Assistants to SEO Data
Screpy hosts the SEO MCP server. Add its address to a compatible assistant and approve the connection in your browser; the assistant can then discover the tools available to your account.
- Use ChatGPT, Claude, Codex, Cursor, Gemini CLI or VS Code with GitHub Copilot through the relevant setup guide; remote HTTP MCP and OAuth support are required.
- Sign in with the account that owns the project or has access to it. A Screpy REST API key does not authorize this connection.
- Ask questions in your assistant’s supported language. Revoke its access in Connected Apps when the connection is no longer needed.
Choose the Right SEO Project
Start by listing your accessible projects, then select the website and evidence you want to investigate. Project names, domains and crawl context help keep an assistant’s answer tied to the right site.
- Choose an owned or shared project returned by Screpy; permissions are checked again whenever a tool runs.
- Select a completed crawl for page-level questions. Ask for the source and date instead of treating every result as a current live check.
- Compare summary metrics for two to five accessible projects, optionally choosing a crawl for each. A project comparison is not a page-by-page difference report.
Investigate Technical SEO Issues
Move from an audit summary to the exact page, link or image behind a finding. Ask your assistant to narrow the completed crawl by structured filters and show the evidence your team needs to fix it.
- Filter crawled pages by recorded conditions such as status codes or missing titles, then open a URL’s stored details.
- Trace a broken link or image back to the pages where it appears; a destination alone does not tell your team where to change the markup.
- Compare two completed crawl summaries and their metric differences. Inspect the selected pages separately when you need to identify individual changes.
- Starting another crawl requires an explicit request, confirmation and available capacity. Processing is asynchronous; inspect completed results after it finishes.
/guides/seo200
Analyze Google Search Console Data
Use the property connected and selected in your Screpy project to investigate Google Search performance. Compare like-for-like periods and keep performance metrics separate from indexing evidence.
- Ask for clicks, impressions, CTR and average position, grouped or filtered by page, query, country or device within the supported query options.
- Compare reporting periods to locate pages or queries worth investigating. Low-volume rows may be omitted, so the returned rows are not a complete search-query export.
- Inspect Google’s indexed view of an eligible URL. URL inspection does not test the live page or submit an indexing request.
- Read the latest existing Index Status audit and its URL findings. MCP cannot start or refresh that audit.
Review Cloudflare SEO Quick Wins
With the matching Cloudflare zone connected in Screpy, your assistant can investigate supported Quick Wins and the signals behind them. Each finding retains its available reporting window and prerequisites.
- Review crawler failures, possible crawl traps, redirect waste and high-transfer URLs where supporting data is available.
- Slow-page opportunities need Web Analytics enabled and enough real-user performance data. Missing evidence does not establish that a page is fast or healthy.
- Use these focused SEO findings alongside crawl and search evidence. They do not expose every analytics report or allow unrestricted Cloudflare setting changes.
linen shirt#11
Impressions
2,860Analyze Tracked Keyword Rankings
Investigate the Rank Tracker results already saved for your project: the keyword, selected market and device, recorded positions and ranking URL. A drop is a reason to investigate, not proof of what caused it.
- Open a tracked keyword’s stored history and results, or filter the keyword list to focus on relevant changes.
- Discover valid tracking options before requesting keyword additions. Add or delete tracked keywords explicitly, with confirmation and capacity checks where required.
- List, preview, create and update URL-rule keyword segments; deletion requires confirmation. A segment preview does not trigger a fresh ranking check.
- Request supported tracking frequency and schedule changes through project settings. MCP does not offer keyword discovery, live SERP lookups, manual reanalysis or Rank Tracker file exports.
Rankings
+24 Position
Investigate AI Search Visibility
See how your brand and competitors appear in the AI Visibility answers stored in Screpy. Open the underlying response and cited sources rather than judging visibility from a mention total alone.
- Explore recorded mentions, competitor visibility, share of voice and sentiment context for the available tracked results.
- Read a saved answer and inspect the prompts, citation sources and recorded prompt or source activity behind a finding.
- Discover supported tracking options, then explicitly request prompt additions or removals. Confirmation and plan capacity still apply where required.
- Request supported analysis-frequency or platform settings through project settings. These tools do not suggest prompts, run live searches, manually rerun analyses or export AI Visibility files.
Mentions
42Analyze Backlinks and Referring Domains
Ask about links to your project domain or a competitor already added in Screpy. First discover valid targets, then move between the profile overview, individual backlinks and the websites that supply them.
- Inspect available source and target URLs, anchors and recorded link status; use report filters to investigate a specific question.
- Group links by referring domain and open the backlinks behind a selected website instead of treating every link as a separate source.
- Compare a backlink gap with one added competitor. Lost and broken backlink-domain reports use the project domain.
- Export supported backlink tables as page-wise CSV or XLSX resources. Follow the returned cursor with the same selectors when you need additional pages.
- Resource reading and file retrieval depend on your AI client’s support and authorized project access. These exports do not include Rank Tracker or AI Visibility files.
- example.org
- example.net
- example.edu
Review and Export Disavows
Use the project’s disavow list to review selected domains or URLs with notes before preparing a file. The list records your review decisions; it does not establish that every selected backlink is harmful.
- List the project’s saved disavow records and review whether an entry applies to a whole domain or one URL.
- Adding up to 100 entries per request or deleting an entry requires explicit confirmation. These actions change only the selected project’s list.
- Export the reviewed entries as authenticated TXT resources, one page at a time. Keep selectors unchanged and join the text pages when preparing the complete file.
- Screpy MCP does not submit the file to Google or guarantee a ranking benefit. Review the completed file and make any submission yourself.
disavow.txtReviewdomain:example.org
https://example.net/links- Domainexample.org
- URLexample.net/links
Check Homepage Performance and Uptime
Add site-health context without confusing a homepage check with site-wide field monitoring. The MCP health tools expose the latest checked homepage and recent stored uptime checks.
- Read available homepage FCP, LCP, CLS, TTFB, TBT and Speed Index measurements. This laboratory snapshot does not provide field INP or performance history for every page.
- Keep an unavailable measurement unavailable. A missing value does not mean zero blocking time or a passing result.
- Review the thirty-day uptime summary and up to 160 newest stored checks, with status, response time and check dates. This is not a complete incident-duration report.
example.comLab
LCP2.1 s
TBT180 ms
CLS0.04
Manage Projects with Your Approval
MCP can perform supported actions inside Screpy when you ask for them. Keep the website, requested change and account capacity clear before allowing your assistant to act.
- Explicitly request project creation or supported settings changes, including crawl and tracking settings. Deleting a project requires confirmation.
- Confirm the exact high-impact action before starting a crawl, removing tracked records or changing disavow entries; an investigation request is not approval for changes.
- Ask for account features and plan limits. The usage tool calculates used and remaining counts for projects; other unavailable usage counters remain null, not zero.
- Feature eligibility, request limits and existing plan capacity still apply. Connecting an assistant does not grant unrestricted access or extra quota.
Review
example.comBuild an SEO Action Plan
Turn a specific investigation into a proposed implementation checklist. Ask your assistant to explain each priority, identify the affected URLs and separate confirmed findings from hypotheses that need another check.
- Tie each task to its crawl, search, ranking, AI or backlink evidence and the relevant analysis date or reporting period.
- Review drafts and recommendations before your team implements them. Screpy MCP does not edit titles, repair website links, publish content or automatically create a scheduled work plan.
- Choose follow-up measurements that match the task: a completed crawl for markup, comparable Search Console periods for search changes and saved tracking results for visibility.
- 1ReviewOpen
/guides - 2SelectedOpen
/shop - 3AfterOpen
example.com
Your AI assistant. Powered by Screpy.
Connect your AI assistant to Screpy through SEO MCP. Ask questions, explore your site data and plan your next steps in the tools you already use.
Connect Screpy through Claude’s remote connector support and investigate completed crawls, exact URLs, and project health.
Connect Screpy to Perplexity with a custom MCP connector. Ask about your site's SEO data on supported Pro, Max, or Enterprise plans.
Authorize Screpy through OAuth in Gemini CLI and use authenticated SEO tools from a terminal-based agent workflow.
Bring Screpy's SEO data into Grok through a custom MCP connector. Explore affected pages, understand findings, and decide what to tackle next.
Connect Screpy to Mistral Vibe with a custom MCP connector. Turn your website data into clear answers and a practical SEO action plan.
Bring crawl, URL, Rank Tracker, and monitoring context into Codex before planning or implementing a code-level SEO change.
Add Screpy as a remote HTTP MCP server and keep crawl evidence beside the code that controls the affected website.
Connect Screpy through VS Code MCP support and use project data from GitHub Copilot Chat while reviewing implementation work.
Connect. Then start talking.
Choose your assistant’s guide, connect your Screpy account through SEO MCP and start with one project. No local MCP server to install.
-
Choose your AI assistant
Open the setup guide for your client. You can keep using the assistant you already know.
-
Connect your Screpy account
Add https://mcp.screpy.com and sign in with Screpy. Approve the connection in your browser; no local server is needed.
-
Ask your first question
Ask “List my Screpy projects”, choose a project and ask what needs attention in its latest completed crawl.
SEO questions you can ask.
Start with the project you want to improve. These SEO MCP questions use the data available in Screpy; connected integrations and saved results determine what your assistant can analyze.
Which technical SEO issues should I fix first?
A prioritized list from your latest completed crawl, with affected URLs and the reasons each issue matters.
Which pages lost Google Search clicks in the last 28 days?
A comparison with the previous 28 days using connected Search Console data, highlighting affected pages and queries.
Which tracked keywords dropped in rankings?
Ranking changes and affected URLs from your saved Rank Tracker history, with pages worth investigating.
What should I investigate in my Cloudflare Quick Wins?
Available Cloudflare-powered findings, their supporting signals and a focused list of follow-up checks.
Where does my brand appear in AI answers?
Brand mentions and citation sources from saved AI Visibility results, with recorded answers you can inspect.
Create a seven-day SEO action plan for this project.
A proposed plan based on available findings, with priorities, affected pages and suggested checks. You review the work before taking action.
Which websites link to my added competitor but not my project?
A backlink gap for one selected added competitor, with referring domains to investigate. It is not an automatic outreach or link-building campaign.
Export the disavow entries I have already reviewed.
Authenticated TXT resource pages from the selected project’s list. Retrieve all requested pages and review the combined file before any manual submission.
You stay in control.
Use the projects you can access, review supported changes and disconnect whenever you choose.
Connect with your account
Sign in through OAuth. Screpy checks access to the selected project whenever a tool runs.
Choose the changes
Ask explicitly for supported changes. Capacity-consuming and destructive actions require confirmation.
Disconnect when you choose
Revoke the client in Connected Apps. Review your AI client’s own data settings before sharing sensitive project data.
SEO tools available through MCP.
Explore the tools your assistant can use. Access depends on your project, connected data and plan; supported changes require your explicit request and confirmation where applicable.
Account and usage
Read account features, plan limits and project usage. Non-project used and remaining counters may be unavailable.
View tool names
-
get_account_context -
get_plan_usage
Projects
List accessible projects, inspect their overview and compare two to five summaries. Create, change or delete projects only when requested, with confirmation where required.
View tool names
-
list_projects -
get_project_overview -
compare_projects -
create_project -
update_project_settings -
delete_project
Website crawls
Select completed crawls, inspect status and summaries, and compare two summary datasets. Starting a crawl requires confirmation and available capacity.
View tool names
-
list_crawls -
get_crawl_status -
start_crawl -
get_crawl_summary -
compare_crawls -
get_on_page_overview
Pages, links and images
Filter stored pages, links and images, inspect URL details and trace their source pages.
View tool names
-
list_pages -
get_page_details -
list_links -
get_link_sources -
list_images -
get_image_sources
Search Console and indexing
Check the selected property, query and compare search performance, inspect indexed URLs and read the latest existing Index Status audit.
View tool names
-
get_search_console_connection -
get_search_console_overview -
query_search_console_performance -
compare_search_console_periods -
inspect_search_console_url -
get_index_status_audit -
list_index_status_urls
Rank Tracker
Inspect saved keyword results and history, and manage tracked keywords when requested. No live SERP lookup or ranking file export.
View tool names
-
get_rank_tracker_options -
get_rank_tracker_overview -
list_rank_tracker_keywords -
get_rank_tracker_keyword -
add_rank_tracker_keywords -
delete_rank_tracker_keywords
AI Visibility
Inspect stored answers, mentions, prompts and sources. Manage monitored prompts when requested; no manual reanalysis or file export.
View tool names
-
get_ai_visibility_options -
get_ai_visibility_overview -
list_ai_visibility_mentions -
get_ai_visibility_answer -
list_ai_visibility_prompts -
get_ai_visibility_prompt_activity -
list_ai_visibility_sources -
get_ai_visibility_source_activity -
add_ai_visibility_prompts -
delete_ai_visibility_prompt
Performance and uptime
Read the latest checked homepage’s laboratory metrics and the thirty-day uptime summary with up to 160 newest checks.
View tool names
-
list_core_web_vitals -
get_uptime_summary
Backlinks and referring domains
Discover project and added-competitor targets, inspect links and domains, compare one-competitor gaps, review lost/broken domains and export page-wise CSV/XLSX resources.
View tool names
-
get_backlink_options -
get_backlinks_overview -
list_backlinks -
list_referring_domains -
list_referring_domain_backlinks -
get_backlink_gap -
list_lost_backlink_domains -
list_broken_backlink_domains -
export_backlink_table
Disavow management
Review the project’s list, add or remove entries with confirmation and export paginated TXT resources for manual review and submission.
View tool names
-
list_disavows -
add_disavows -
delete_disavow -
export_disavows
Quick Wins
Find SEO opportunities from search performance, completed crawls and connected traffic data. Filter by category to explore the supporting findings.
View tool names
-
list_quick_wins
Keyword segments
Group tracked keywords with keyword or URL rules. Preview matches and average positions before saving; create, update or delete segments when requested.
View tool names
-
list_rank_tracker_segments -
preview_rank_tracker_segment -
create_rank_tracker_segment -
update_rank_tracker_segment -
delete_rank_tracker_segment
SEO MCP questions, answered.
How to connect, what you can ask and how Screpy MCP uses your website data.
What is an SEO MCP server?
An SEO MCP server connects a compatible AI assistant to SEO tools and data through the Model Context Protocol. Screpy’s hosted server exposes 64 project-scoped tools for accessible projects, crawl evidence, Search Console, stored rankings and AI results, backlinks, disavows and health data. The assistant uses those tools during a conversation; available data and supported actions still depend on the account and project.
Which AI assistants support Screpy MCP?
Screpy has setup guides for ChatGPT, Claude, Codex, Cursor, Gemini CLI and VS Code with GitHub Copilot. The client must support remote HTTP MCP connections and browser-based OAuth. Availability depends on the client’s plan, workspace permissions and interface; check its guide before connecting.
Do I need coding skills to use Screpy MCP?
No. Once connected, describe what you want in everyday language: “What should I fix first on example.com?” Your assistant can find the project and relevant data. You do not need to write API requests or look up project IDs.
Can I use SEO MCP in my language or by voice?
Ask in a language your AI assistant supports and request a simpler explanation when needed. Voice use depends on whether that client supports the Screpy connection in its voice interface. MCP supplies the data; the assistant provides the language and voice experience.
Which plan and usage limits apply to Screpy MCP?
Your Screpy plan determines feature eligibility and capacity. Starting a crawl or adding tracked keywords or AI prompts does not bypass existing limits. Ask for account features and plan limits; the usage tool reports used and remaining counts for projects, while other usage counters may be unavailable. An unavailable counter is not zero. Your AI client may also require its own subscription or workspace approval.
How do I disconnect an AI assistant from Screpy?
Open Connected Apps in Screpy, find the authorized AI client and revoke its connection. Further MCP access requires a new OAuth approval. Revocation does not delete results already stored by the AI client; manage those through its own data controls.
What Cloudflare data can I use through Screpy MCP?
With Cloudflare connected to the matching site zone in Screpy, you can explore supported Quick Wins for crawler failures, possible crawl traps, redirect waste and high-transfer URLs. Slow-page findings need Web Analytics enabled and enough performance data. This provides specific SEO findings, rather than unrestricted access to every Cloudflare report or setting.
Why can’t my assistant find my project or SEO data?
First ask “List my Screpy projects” and check that you signed in to the right account. The project must be yours or shared with you. Missing results can also mean no completed crawl, disconnected integrations or no tracking history. If authorization fails, reconnect; reconnecting cannot create missing analysis data.
Can Screpy MCP show my brand’s AI search visibility?
Yes, from AI Visibility results saved in your project. Inspect brand mentions, competitor visibility, share of voice, recorded answers and citation sources. Explicitly request additions or removals of monitored prompts, with confirmation where required. These tools do not export AI Visibility files, suggest prompts or manually rerun analyses.
Can my AI assistant download backlink and disavow files?
The export tools return authenticated MCP resource links for supported backlink CSV/XLSX tables and disavow TXT entries. Each export covers one page; follow its next cursor with unchanged selectors to retrieve additional pages, and join TXT pages for a complete disavow file. Your client must support reading those resources. These are not anonymous public download URLs, and Rank Tracker or AI Visibility file exports are not available through MCP.
Does Screpy MCP fix or publish website changes automatically?
No. It retrieves evidence and can help your assistant draft recommendations, but it does not edit or publish the website, submit indexing requests or send disavow files to Google. Supported actions inside Screpy need an explicit request; high-impact actions also require confirmation. A proposed action plan is not an automatically created or scheduled set of tasks.
How do I connect Screpy MCP?
Add https://mcp.screpy.com in your assistant’s MCP or connector settings, sign in to Screpy and approve the connection in your browser. Then ask “List my Screpy projects.” Screpy hosts the server, so no local installation is needed. Use OAuth, not a Screpy REST API key.
Who can access my SEO data through MCP?
The authorized assistant can request projects you own or that are shared with your Screpy account. Screpy checks access on every tool call. Data returned to the AI client may be processed or retained under that client’s own settings and terms. Review its privacy controls before connecting sensitive projects.
Is the SEO data returned by MCP real time?
Not all of it. MCP retrieves available data when you ask, but crawls, rankings and AI visibility reflect completed or saved analyses. Search Console and Cloudflare have their own reporting delays and data windows. Ask for the source, analysis date and reporting period before drawing conclusions.
What do I need before asking about my website?
You need a Screpy account, access to the project and a connected compatible assistant. Website audits need completed crawl data; Search Console needs a connected property; rankings and AI visibility need saved tracking results. Connect Cloudflare in Screpy for supported traffic findings.
Can I check keyword rankings through Screpy MCP?
Yes. Your assistant can review Rank Tracker keywords, saved position history, ranking URLs and keyword segments. Ask which tracked keywords gained or lost positions. You can also explicitly request keyword additions or deletions, with confirmation where required. MCP does not provide live SERP lookups or manual keyword reanalysis.
Can I analyze Google Search Console data with MCP?
Yes, when your Screpy project has a connected and selected Search Console property. Ask about clicks, impressions, CTR or average position, compare periods, and filter by page, query, country or device. You can also inspect a URL’s indexed view and existing Index Status findings. Inspection does not request indexing or test the live page. Search Console may omit low-volume query rows.
Can I investigate technical SEO issues through MCP?
Yes. Select a completed crawl to filter pages, links and images, open stored URL details and trace a finding to its source page. Compare completed crawl summaries when you need metric differences. The separate health tools provide the latest checked homepage and recent uptime context. A new crawl requires an explicit request, confirmation and available capacity; Index Status tools cannot start a new indexing audit.
Does SEO MCP report Core Web Vitals for every page?
No. The health tool reads the latest checked homepage’s available FCP, LCP, CLS, TTFB, TBT and Speed Index values. It is a laboratory snapshot, not a field INP report or full-site performance history. Separately, Cloudflare-powered Quick Wins can expose supported slow-page findings when Web Analytics and enough real-user data are available; the two sources should not be merged into one measurement.
How is SEO MCP different from asking AI for SEO advice?
An unconnected assistant cannot retrieve your private Screpy project data. With Screpy MCP connected, it can check available findings before answering, show affected URLs and help prepare an SEO action plan. Ask for sources and dates, and review recommendations before acting.
What is the difference between Screpy MCP and the SEO API?
Use MCP to let a compatible AI assistant discover and call Screpy tools during a conversation. Use the Screpy REST SEO API for your own products, dashboards, scripts or automations using direct HTTP requests and API keys. Their authentication and available capabilities are different.
Does Screpy MCP support backlink analysis and keyword research?
Backlink analysis is supported: discover valid project or added-competitor targets, inspect backlinks and referring domains, compare one added competitor’s backlink gap, and review the project’s lost or broken backlink domains. Supported tables can be exported as paginated CSV/XLSX resources. Keyword-discovery tools are not exposed; MCP reads and manages tracked Rank Tracker keywords instead. Use the dashboard’s dedicated Keyword Research feature for discovery.
Go further with your SEO workflow.
Start with the connection guide, then explore practical examples and the data behind your questions.
Start with one question.
Connect your AI assistant to Screpy and ask: “What should I focus on first?”