> ## Documentation Index
> Fetch the complete documentation index at: https://praison.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Firecrawl PraisonAI Integration
> Guide for integrating Firecrawl web scraping capabilities with PraisonAI, including custom tool implementation for web page content extraction
Give an Agent a Firecrawl scraping tool so it can read and summarise any web page.
```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
from praisonaiagents import Agent
def scrape_page(url: str) -> str:
"""Scrape and return the main content of a web page."""
from firecrawl import FirecrawlApp
app = FirecrawlApp(api_url="http://localhost:3002")
return app.scrape_url(url=url)["content"]
agent = Agent(instructions="Summarise web pages", tools=[scrape_page])
agent.start("Summarise https://praison.ai")
```
```mermaid theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
graph LR
subgraph "Firecrawl Scraping"
User[📋 User] --> Agent[🤖 Agent]
Agent --> Firecrawl[🔧 Firecrawl]
Firecrawl --> Result[✅ Page Content]
end
classDef input fill:#8B0000,stroke:#7C90A0,color:#fff
classDef process fill:#189AB4,stroke:#7C90A0,color:#fff
classDef output fill:#10B981,stroke:#7C90A0,color:#fff
class User,Agent input
class Firecrawl process
class Result output
```
# Firecrawl PraisonAI Integration
## Firecrawl running in Localhost:3002
```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
from firecrawl import FirecrawlApp
from praisonai_tools import BaseTool
import re
class WebPageScraperTool(BaseTool):
name: str = "Web Page Scraper Tool"
description: str = "Scrape and extract information from a given web page URL."
def _run(self, url: str) -> str:
app = FirecrawlApp(api_url='http://localhost:3002')
response = app.scrape_url(url=url)
content = response["content"]
# Remove all content above the line "========================================================"
if "========================================================" in content:
content = content.split("========================================================", 1)[1]
# Remove all menu items and similar patterns
content = re.sub(r'\*\s+\[.*?\]\(.*?\)', '', content)
content = re.sub(r'\[Skip to the content\]\(.*?\)', '', content)
content = re.sub(r'\[.*?\]\(.*?\)', '', content)
content = re.sub(r'\s*Menu\s*', '', content)
content = re.sub(r'\s*Search\s*', '', content)
content = re.sub(r'Categories\s*', '', content)
# Remove all URLs
content = re.sub(r'http\S+', '', content)
# Remove empty lines or lines with only whitespace
content = '\n'.join([line for line in content.split('\n') if line.strip()])
# Limit to the first 1000 words
words = content.split()
if len(words) > 1000:
content = ' '.join(words[:1000])
return content
```
## How It Works
The Agent calls the Firecrawl tool with a URL, Firecrawl fetches and cleans the page, and the Agent summarises the result.
```mermaid theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
sequenceDiagram
participant User
participant Agent
participant Firecrawl
User->>Agent: Summarise this URL
Agent->>Firecrawl: scrape_url(url)
Firecrawl-->>Agent: Cleaned page content
Agent-->>User: Summary
```
## Best Practices
Point `FirecrawlApp(api_url="http://localhost:3002")` at a local Firecrawl instance to keep scraped data on your machine.
Strip menus, URLs, and boilerplate and cap the length so the model sees only the relevant text.
Add a type hint and docstring so the Agent builds the correct tool schema for the model.
## Related
Learn how agents call tools.
Build your first agent.