Skip to main content
skills.sh The Geekflare agent skill teaches an AI agent when to reach for Geekflare and how to call each tool well, so it can answer from public web data instead of guessing. It covers:
  • Search, including full Google SERP results
  • Web scraping into Markdown or structured JSON, and page metadata
  • Brand identity (logos, colors, fonts)
  • Screenshots and URL-to-PDF
  • Website diagnostics such as uptime, DNS, TLS/SSL, redirects, broken links, load time, and Lighthouse
The skill is open source: github.com/geekflare/geekflare-skills.

​
Install

Works with Claude Code, Cursor, Codex, GitHub Copilot, Windsurf, Gemini and other agents that support skills.
See the listing on skills.sh.
The skill works best with the Geekflare MCP server connected. When the MCP tools are available, the agent uses them directly. Set up your client first using one of the client guides. If MCP isn’t connected, the skill falls back to calling the REST API with curl, as long as the agent can run shell commands and your API key is set.

​
Set your API key

The skill reads your key from the GEEKFLARE_API_KEY environment variable, so you don’t need to paste it into chat or commit it to a config file.
Replace YOUR_API_KEY with your Geekflare API key. Get one at dash.geekflare.com. Where your client expands environment variables in its MCP config, you can reference the variable instead of the key:
Add this to .mcp.json:
Each Geekflare call uses credits from your account. The skill is designed to avoid spending credits on requests that a cheaper tool, or no tool, can answer.

​
Next Steps

MCP Server

Connect Geekflare tools to your AI client.

Integrations

Zapier, Make, n8n, Composio, SDKs and more.