Investigate a traffic dip
Open Claude Desktop, ask why pageviews fell on /pricing last week. Claude reads the slice through MCP, returns the directional answer.
Connect Claude Desktop, Claude Code, or the VS Code Claude extension to WebmasterID. The integration is workspace-scoped, read-only, and audit-logged by construction.
The use cases below are the day-to-day shape of the workflow. Each one is a prompt the operator runs alongside the dashboard, not instead of it.
Open Claude Desktop, ask why pageviews fell on /pricing last week. Claude reads the slice through MCP, returns the directional answer.
“Summarise the AI crawlers and AI assistant referrals this week.” One prompt, one summary, audit-logged.
Claude Code reads the WebmasterID Agent recommendations, opens the file in the repo mapping, prepares a diff. You review and ship.
Drop the conversation into a doc. The data is reproducible because Claude fetched it through MCP, not from a screenshot.
Ask Claude to read the recent events on a newly-configured site. Confirm the tracker is firing without opening the dashboard.
Multi-site Claude prompts work because the MCP key is workspace-scoped — one connection sees the whole portfolio.
The MCP server is the integration layer. Each Claude client is just a different way of talking to it.
The conversational interface for day-to-day reading and summarisation. Paste the MCP config snippet, restart the app, and the WebmasterID tools appear in the available-tools list.
The CLI interface for pairing analytics reads with code changes. Read Agent recommendations, open the file in the repo mapping, prepare a diff. Operator reviews and ships.
The editor-side interface. Useful when an investigation begins from a file or repository rather than a dashboard.
Safety model. The MCP surface is read-only. Operator approval is required for every change — there is no auto-deploy, no auto-merge, no hidden execution.
A small, deterministic surface. The same tools Claude sees are documented in the dashboard so operators always know what was read.
The Event Explorer surface mirrored for AI clients. Filterable by date, type, country, pathname, referrer, session, visitor.
AI crawler timeline and AI assistant referrals. Useful for AI-search-era investigations.
Directional source attribution with confidence labels. Claude can read and summarise; the operator decides.
WebmasterID Agent recommendations and the daily operator brief. Helpful for AI-assisted prioritisation.
Operators rightly worry about silent data exfiltration when AI clients connect to internal data. WebmasterID's Claude integration is built to remove that worry.
Workspace-scoped key
The Bearer key Claude uses is tied to a single workspace. Cross-workspace reads are impossible.
Read-only
The MCP surface has no write tools. Claude can read; the operator deploys.
Audit-logged
Each read appends to the audit log. You can see exactly what Claude requested, when, and by which key.
Revocable
Delete the key to terminate access. The audit log is preserved so you keep a complete history.
Questions about the Claude integration? info@helperg.com.
The pieces the Claude integration depends on.
MCP analytics
The Model Context Protocol surface that powers the Claude integration.
Read more →
AI-assisted analytics
The product story around the AI-assisted workflow shape.
Read more →
AI visibility analytics
The first surface Claude reads — AI crawlers and AI-assistant referrals.
Read more →
Event Explorer
The investigation surface mirrored for AI clients.
Read more →
Multi-site analytics
Workspace-scoped MCP is the agency workflow shape.
Read more →
Docs
Configure the Claude integration end-to-end.
Read more →