What is the krawlup MCP server?

A Model Context Protocol endpoint that lets AI agents such as Cursor, Claude Code, Codex, ZCode, Command Code, OpenCode, Windsurf, Zed and GitHub Copilot read your krawlup scores, failing audits, AI fix cards, scan history and first-party analytics, start full rescans, and manage your directory presence. One endpoint, one browser approval.

How do I connect Cursor, Claude Code, ZCode or Command Code?

Pick your client in the Connect section and approve the connection in your browser. No API key needed, and the connection sticks. Prefer keys for scripts or headless agents? Create one on the AI agent page in your dashboard and send it as a bearer token.

Yes. Connections and agent keys are free on every plan, and the read tools are unlimited within the rate limits. trigger_rescan spends the same plan allowances as clicking Rescan site in your dashboard, and publishing to the public directory is paid-only.

What can the agent change?

With the submit permission it can add a product in your name and publish it to the public directory. Adding always starts private, publishing is paid-only, and the agent never spends money. Every payment stays in the browser.

How do I revoke access?

Connections and keys both live on the AI agent page in your dashboard. Disconnecting or deleting stops the agent on its very next request. Keys can also be restricted to specific products and given an expiry date.

krawlup MCP server

Your growth data, inside your AI agent.

Connect Cursor, Claude Code, Codex, ZCode, Command Code, Windsurf, Zed or any MCP client. Your agent reads scores, fix cards, analytics and keyword ideas, then rescans, submits and publishes on your behalf.

Connect your agent Browse the tools

Works with Claude Code, Codex, Cursor, Claude, ZCode, Command Code, OpenCode, GitHub Copilot, Cline and any MCP-compatible client.

Try asking your agent

Fix

“Check my krawlup scores, find the biggest issue, and fix it in this repo.”

get_insightsget_fix_cards

Measure

“Did last week's fix actually move traffic and conversions?”

get_scan_historyget_website_analytics

Compare

“Compare my products by revenue and visitors for the last 30 days and tell me which one needs attention.”

compare_productsget_my_subscription

Research

“Run a free krawlup site report on example.com and tell me the three highest-impact fixes.”

get_site_reportrun_site_report

Ship

“Add myproduct.com to krawlup and make it public.”

check_launch_readinessadd_productset_product_public

Connect a website first. Your agent only sees what your credential allows, and catalog searches are public to every agent.

Connect your assistant

Add krawlup as a remote MCP server. Both methods use the same URL over streamable HTTP, and the tool list follows your credential and plan.

MCP URLhttps://api.krawlup.com/mcp

Claude Code connects over streamable HTTP with one command; the browser handles the approval.

1Run this in your project root.
2When the agent first needs krawlup it opens your browser; approve while signed in.
3The connection sticks. Manage or remove it on the Dashboard → AI agent page.

Terminal

claude mcp add --transport http krawlup https://api.krawlup.com/mcp

Windsurf, Zed and VS Code accept the same mcpServers JSON shape as Cursor. Scopes, limits and every tool: krawlup.com/docs.

From one question to your daily workflow

Sixteen tools over one endpoint. The catalog reads are public; everything personal needs your credential.

Plan awareness

The agent checks your plan first: status, abilities, limits and this period’s usage meters — so it never wastes a call on something your subscription doesn’t include.

get_my_subscription

Portfolio compare

All your products side by side, sorted by visitors, views, revenue totals, average revenue or score — lifetime or over a window. Visitor and revenue columns follow the same subscription gating as the dashboard.

compare_products

Audits and insights

Performance, AEO and authority scores with the failing audits behind them, plus the one edit that matters most right now.

get_insightsget_bottleneck

AI fix cards

What broke, why it matters, and a paste-ready prompt your agent can implement in the repo.

get_fix_cards

First-party analytics

Visitors, sessions, referrers and conversion goals from the cookieless tag, plus the snippet to install it.

get_website_analyticsget_analytics_snippet

Keyword ideas

Intent-clustered keywords with volume and difficulty estimates and prioritized content actions.

get_keyword_ideas

Free site report

The public site-report audit for any website, no sign-up: performance, AEO, CRO, SEO crawl and the named checks. Two full scans per agent per month; reads unlimited.

get_site_reportrun_site_report

Directory presence

Audit a URL, add the product from just its URL, and publish it to the public directory on paid plans.

add_productset_product_public

Rescans

Start a full re-analysis after shipping a fix and poll for the new scores. Keys only; spends plan allowances.

trigger_rescan

Read the complete tool reference

You choose the access

Every tool call checks the credential, its scopes and product access before anything moves.

Read-only by default

Keys start at the read scope. Rescan and submit are opt-in checkboxes you flip at creation.

OAuth is scoped

A browser-approved connection gets read and submit. It is never granted rescan.

Revoke in one click

Connections and keys live on the AI agent page. Disconnecting stops the agent on its next request.

Money stays yours

The agent never touches payments. Publishing is paid-only, and billing happens in the browser.

Manage everything on the AI agent page in your dashboard.

Questions founders ask

Building an integration?

Read the docs llms.txt llms-full.txt

Official packages on npm

@krawlup/skills @krawlup/analytics-tag

Cookieless, GDPR-friendly and scoped: agents only ever see what your credential allows.