Web Fetch
Retrieve and analyze URL content (Anthropic only)
The --web-fetch flag enables URL content retrieval and analysis using Anthropic’s native web fetch capability.
Quick Start
praisonai "Summarize https://praison.ai/docs" --web-fetch --llm anthropic/claude-sonnet-4-20250514
Usage
Basic Web Fetch
praisonai "Summarize https://praison.ai/docs" --web-fetch --llm anthropic/claude-sonnet-4-20250514Expected Output:
📥 Web Fetch enabled
╭─ Agent Info ─────────────────────────────────────────────────────────────────╮
│ 👤 Agent: DirectAgent │
│ Role: Assistant │
│ Model: anthropic/claude-sonnet-4-20250514 │
│ Web Fetch: Enabled │
╰──────────────────────────────────────────────────────────────────────────────╯
╭─ Fetching URL ───────────────────────────────────────────────────────────────╮
│ 🔗 https://praison.ai/docs │
╰──────────────────────────────────────────────────────────────────────────────╯
╭────────────────────────────────── Response ──────────────────────────────────╮
│ # Summary of PraisonAI Documentation │
│ │
│ PraisonAI is a production-ready Multi-AI Agents framework with: │
│ - Self-reflection capabilities │
│ - Integration with CrewAI and AutoGen │
│ - Low-code solution for multi-agent systems │
│ ... │
╰──────────────────────────────────────────────────────────────────────────────╯Combine with Other Flags
# Web fetch with save
praisonai "Analyze https://example.com/article" --web-fetch --save --llm anthropic/claude-sonnet-4-20250514
# Web fetch with metrics
praisonai "Extract data from https://api.example.com" --web-fetch --metrics --llm anthropic/claude-sonnet-4-20250514Requirements
| Requirement | Value |
|---|---|
| Provider | Anthropic only |
| Models | claude-sonnet-4-20250514, claude-3-opus, claude-3-sonnet, claude-3-haiku |
| API Key | ANTHROPIC_API_KEY environment variable |
YAML Equivalent
Use web_fetch: true on an agent with an Anthropic model:
agents:
fetcher:
role: Web Content Analyzer
instructions: "Fetch and analyze web content"
web_fetch: true
llm: anthropic/claude-sonnet-4-20250514How It Works
- Enable: The
--web-fetchflag activates Anthropic’s URL fetching - Fetch: Claude fetches the content from the specified URL
- Parse: Content is parsed and cleaned
- Analyze: Claude analyzes the content based on your prompt
- Respond: Synthesized response is returned
Use Cases
Summarization
praisonai "Summarize this article: https://example.com/article" \
--web-fetch --llm anthropic/claude-sonnet-4-20250514Data Extraction
praisonai "Extract all product prices from https://example.com/products" \
--web-fetch --llm anthropic/claude-sonnet-4-20250514Content Analysis
praisonai "Analyze the sentiment of https://example.com/review" \
--web-fetch --llm anthropic/claude-sonnet-4-20250514Documentation Review
praisonai "Review the API documentation at https://api.example.com/docs" \
--web-fetch --llm anthropic/claude-sonnet-4-20250514Programmatic Usage
from praisonaiagents import Agent
agent = Agent(
instructions="You are a content analyzer",
llm="anthropic/claude-sonnet-4-20250514",
web=True
)
result = agent.start("Summarize https://praison.ai/docs")
print(result)Comparison with Web Search
| Feature | Web Fetch | Web Search |
|---|---|---|
| Purpose | Fetch specific URL | Search the web |
| Input | URL required | Query |
| Depth | Full page content | Search snippets |
| Provider | Anthropic only | Multiple |
Best Practices
| Do | Don’t |
|---|---|
| Use for specific URLs | Use for general search |
| Check URL accessibility | Assume all sites work |
| Use for content analysis | Use for real-time data |
| Combine with save for long content | Rely on terminal output |