An advanced AI-powered web scraping and research skill that enables autonomous browsing, structured data extraction, and deep web mapping.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install firecrawl-mcp
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install firecrawl-mcp using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is firecrawl-mcp?
The firecrawl-mcp skill is a high-performance toolset designed for Openclaw Skills users who need to transform the web into structured data. By leveraging the OneKey Gateway, it provides a unified interface for complex web operations, ranging from simple URL scraping to deploying autonomous research agents that can navigate multiple sites to find specific answers.
This skill excels at handling modern, JavaScript-heavy websites (SPAs) and offers unique features like brand identity extraction and persistent browser sessions via Chrome DevTools Protocol (CDP). Whether you need to monitor competitors, gather training data for LLMs, or automate multi-step browser workflows, firecrawl-mcp provides the technical infrastructure to do so reliably and at scale.
firecrawl-mcp Use Cases
- Extracting specific structured data like pricing, API specifications, or product details using JSON schemas.
- Conducting autonomous market research across multiple domains to find startups or investment opportunities.
- Mapping entire website architectures to discover hidden endpoints or documentation pages.
- Replicating design systems by extracting brand identity profiles including colors, typography, and UI components.
- Automating complex, multi-step browser interactions such as clicking, scrolling, and form submission via persistent CDP sessions.
How firecrawl-mcp Works
- Authentication: The skill initializes via the OneKey Gateway using the DEEPNLP_ONEKEY_ROUTER_ACCESS API key.
- Discovery: Users can use the map or search tools to identify target URLs or find relevant information across the web.
- Extraction Strategy: Depending on the need, the system selects between simple scraping (markdown/JSON), crawling (multi-page), or autonomous agent research.
- Dynamic Rendering: For JavaScript-heavy sites, the skill utilizes 'waitFor' parameters or persistent browser sessions to ensure content is fully loaded.
- Data Structuring: Raw web content is processed through LLM-backed extraction to fit a predefined JSON schema or design profile.
- Async Retrieval: For long-running research tasks, a Job ID is issued which the user polls until the process reaches a completed status.
firecrawl-mcp Setup
First, ensure you have the necessary CLI and package dependencies installed:
npm -g install @aiagenta2z/onekey-gateway
pip install ai-agent-marketplace
Set your OneKey access key as an environment variable to authenticate with Openclaw Skills:
export DEEPNLP_ONEKEY_ROUTER_ACCESS=YOUR_API_KEY
To run a basic scrape via the CLI:
npx onekey agent firecrawl-mcp/firecrawl-mcp firecrawl_scrape '{"url": "https://example.com", "formats": ["markdown"]}'
firecrawl-mcp Data Schema & Taxonomy
The firecrawl-mcp skill organizes data based on the requested format and tool used:
| Tool | Primary Output | Metadata Taxonomy |
|---|---|---|
| firecrawl_scrape | Markdown, JSON, or Branding | URL, status, cached status, extraction schema |
| firecrawl_map | Array of Strings | List of discovered internal/external URLs |
| firecrawl_search | Array of Objects | Title, URL, snippet, and optional scraped content |
| firecrawl_extract | Structured JSON | Matches user-defined JSON schema (properties, types) |
| firecrawl_agent | Job Object | ID, status (processing/completed/failed), research findings |
firecrawl-mcp Advanced Features
- Autonomous Research Agent: A multi-step AI layer that browses and navigates the web independently to answer complex queries.
- Persistent Browser Sessions: Create and manage browser instances using firecrawl_browser_create for multi-call automation.
- CDP Execution: Run custom Python, Node.js, or Bash (agent-browser) scripts directly within a live browser environment.
- Brand Identity Extraction: A specialized format that analyzes CSS and assets to return a comprehensive design profile.
- Performance Caching: Use the maxAge parameter to achieve significantly faster response times (up to 500% faster) via cached data.
SKILL.md
Loading
Related Openclaw Skills
RPM Packaging Standard
A definitive technical guide and automation resource for creating, building, and maintaining high-quality RPM packages across all major Linux distributions.
Xiaohongshu Keyword Collector
An automated browser-based tool that extracts real-time search suggestions and trending keywords directly from Xiaohongshu's Explore page.
Promotion Engine
A smart retail promotion calculator that computes final prices, applies discount stacking rules, and provides transparent savings breakdowns for customers.
Multi-Environment Isolator
A professional tool to generate fully isolated development, testing, and production environments for Python web projects with integrated frontend support.
Megaventory
A comprehensive AI agent integration for Megaventory to automate inventory tracking, order processing, and supplier management.
Contest Page Generator
A technical framework for designing, auditing, and optimizing high-conversion contest, giveaway, and promotional campaign landing pages.
Featured*
