Agents & API (MCP)
Let your own AI agents run audits and read the results: Claude Code, Cursor, or anything that speaks MCP.
A built-in MCP server lets agents like Claude Code list your sites, start audits, wait for them and read scores, issues, findings and recommendations, with a personal API token that also works on the REST API.
- 24 tools: audits, issues, recommendations, rankings, keywords and more
- One command to connect Claude Code
- Read-only or read-and-run tokens, pinned to one organization
Use cases
What you can do with it
Developers
SEO from your editor
Ask Claude Code to audit the staging site, wait for it, and fix the broken links and missing canonicals it finds, right in the codebase.
Engineering teams
Automated checks
Start an audit after a deploy from a script and fail the pipeline when critical issues appear.
Agencies
Your own reporting
Pull scores, issues and recommendations into your own dashboards or AI workflows.
Why it's strong
Built to be trusted
Standard MCP
Streamable HTTP at /api/mcp, so any MCP client can connect. Claude Code takes one command.
Built for agents
A wait tool reports progress while an audit runs, and list tools page their results so an agent's context stays small.
Scoped tokens
Each token works in one organization only, can be read-only, expires, and can be revoked at any time.
People still decide
Approving or rejecting recommendations and adding sites stay in the web app, with a person.
How it works
From setup to results
- 1
Create a token
Settings → API & MCP: name it, pick access and expiry.
- 2
Connect your agent
Paste the claude mcp add command shown with the token.
- 3
Ask
"Audit example.com, wait for it, and summarise the top issues."
Data it uses
- Everything in your organization that the token's user can see
FAQ
Good to know
Works well with
All featuresAI SEO analyst
Not a prompt over a findings list: an analyst that decides what to look at, digs in, and checks its own answer.
Learn moreTechnical SEO audit
Crawl your site like a search engine and see exactly what holds it back.
Learn moreRecommendations you approve
A short list of fixes, each backed by evidence, and nothing changes until you say so.
Learn more