# Searchkeep

> Self-hosted SEO & GEO analytics that runs in your own Cloudflare account. Connect Google Search Console, Bing Webmaster Tools and Cloudflare, see which pages AI crawlers read, cite and send clicks to, and ask about it over MCP.

Searchkeep is open source (MIT) and self-hosted: one Cloudflare Worker plus one D1 database in the user's own Cloudflare account. It analyzes only the user's own sites using first-party data (Google Search Console, Bing Webmaster Tools, Cloudflare AI crawler traffic and Web Analytics). It exposes an MCP server at `https://<your-domain>/mcp` with 12 read-only tools (plus `submit_sitemap`, registered only when writes are enabled) and a dashboard. It does not do backlinks, competitor research or rank tracking, by design. There is no hosted version and no telemetry.

- Website: https://searchkeep.dev/
- Source code: https://github.com/noobnooc/searchkeep
- Current version: v0.1.0 (released 2026-11-03)
- Full documentation in one file: https://searchkeep.dev/llms-full.txt

## Getting started

- [Introduction](https://searchkeep.dev/docs): What Searchkeep is, how it works, what it deliberately does not do, and what you need to run it in your own Cloudflare account.
- [Deploy](https://searchkeep.dev/docs/deploy): Step-by-step guide to deploying Searchkeep to your Cloudflare account with a GitHub fork and Workers Builds, including Access, credentials and verification.
- [Configuration](https://searchkeep.dev/docs/configuration): Every Worker Secret Searchkeep reads, the SITES JSON format, brand regex, write access, sync budgets, how to set secrets with wrangler or the dashboard, and local development.

## Using Searchkeep

- [MCP server](https://searchkeep.dev/docs/mcp): Connect Claude, Claude Code, Cursor, ChatGPT or any MCP client to your Searchkeep endpoint, and the full reference for every tool, its parameters and defaults.
- [Data sources](https://searchkeep.dev/docs/data-sources): What Searchkeep pulls from Google Search Console, Bing Webmaster Tools and Cloudflare, the credentials each needs, how syncing works, how AI citations are imported and every known limitation.
- [Security](https://searchkeep.dev/docs/security): How Searchkeep is protected: Cloudflare Access, Managed OAuth for MCP clients, JWT re-verification on every request, readiness states and locking until ready.

## Project

- [Architecture](https://searchkeep.dev/docs/architecture): One Worker, three entrances: how MCP, the dashboard and the cron share one core and one D1 database, the package layout and the extension points.
- [Upgrading](https://searchkeep.dev/docs/upgrading): Upgrade Searchkeep by syncing your fork. What happens during an upgrade, how migrations work, manual steps and rolling back.
- [FAQ](https://searchkeep.dev/docs/faq): Frequently asked questions about Searchkeep: cost, data ownership, GEO, history, AI clients and upgrades.

## Optional

- [Changelog](https://searchkeep.dev/changelog): Release notes and upgrade notes.
- [Privacy](https://searchkeep.dev/privacy): The website sets no cookies and uses no trackers; the product sends no telemetry.
- [GitHub repository](https://github.com/noobnooc/searchkeep): Source, issues, AGENTS.md and docs/deploy.md (the agent-executable deploy runbook).