SEO MCPSEO MCP Workflows
Choose an SEO MCP workflow for technical audits, broken links, crawl changes, Rank Tracker, AI Visibility, and page experience.
An SEO MCP workflow starts with a decision, not a prompt collection. Choose the question you need to answer, check which Screpy data already exists, and follow the linked procedure to a bounded, verifiable result. The server reads data from projects your signed-in account can access; it does not make a missing crawl or analysis appear.
Choose the evidence you need
| Your question | Start with | Required evidence | What the answer should contain |
|---|---|---|---|
| Which technical issues affect important pages? | Technical SEO audit | A completed crawl | Affected URLs, issue fields, likely impact, and validation steps |
| Where are broken links and where do they appear? | Broken links | Link records from a completed crawl | Destination, source pages, anchor context, and priority |
| Which images need attention? | Image SEO | Image records from a completed crawl | Asset, source pages, reason, and fix |
| Did a release resolve or create issues? | Compare crawls | Two comparable completed crawls | Changed metrics, representative URLs, and uncertainty |
| Which pages have weak experience signals? | Core Web Vitals | Available field-performance records | Metric, page group, measurement scope, and next diagnostic |
| What is the most useful next SEO task? | Quick Wins | Available search, crawl, or connected traffic data | Source, opportunity, verification, and effort |
| Which tracked queries moved? | Rank Tracker analysis | Stored keyword analyses | Market, device, date, position history, and matching URL |
| Where does an AI answer mention us? | AI Visibility investigation | Stored prompts and answers | Provider, answer, brand mention, citation, and limitations |
The evidence sources differ. A crawler record is not a Search Console query, a stored SERP result is not a live Google lookup, and an AI answer is a dated observation rather than a stable ranking. Keep those distinctions in the final answer.
A reliable investigation path
- Use
list_projectsand select the returned project UID. Check whether the project belongs to the signed-in member or is shared with them. - Discover the data source for the chosen workflow. Use
list_crawlsbefore crawl-level tools; use the Rank Tracker or AI Visibility read tools for their own stored datasets. - Begin with a summary, then retrieve a bounded set of exact records. Follow cursors when the answer depends on more than the first page.
- Separate observed facts, possible causes, and the change you recommend. Name the filter, date, device, and URL when they affect interpretation.
- Verify a fix with a later comparable crawl or analysis. Do not call a write tool unless the user requested that action; capacity-consuming or destructive changes require the tool's confirmation step.
If the needed data is missing
Start with connection and project access, then check troubleshooting. A completed crawl is needed for page, link, and image evidence. Rank Tracker and AI Visibility require their own tracked keywords or prompts and completed stored analyses. Report unavailable data rather than presenting an empty list as a healthy result.
For the available tool names and limits, use the tools reference. For prompt wording after selecting a workflow, use the prompt library.
Related documentation
Continue exploring SEO MCP
Run Your First SEO Audit with SEO MCP
Follow a practical SEO MCP workflow that finds high-impact technical SEO issues and verifies every recommendation against crawl records.
Technical SEO Audit Workflow
Use SEO MCP to run a structured technical SEO audit across crawlability, indexing signals, metadata, content structure, links, images, and performance.