Available Tools

InfluencerPy agents have access to several powerful tools for content discovery and analysis. Each tool serves a specific purpose and can be combined to create sophisticated content workflows.

Core Tools

1. HTTP Request Tool 🌐

Purpose: Fetch and parse content from any web URL using Beautiful Soup.

Best for: - Reading blog posts and articles - Extracting specific content using CSS selectors - Monitoring webpage content - Getting links from pages

Usage:

http_request(
    url="https://example.com/article",
    selector="article",  # Optional: CSS selector
    extract_links=True   # Optional: Extract all links
)

Features: - Clean text extraction (removes scripts, styles) - CSS selector support for targeting specific elements - Link extraction with absolute URLs - Automatic content truncation (10,000 char limit) - 10-second timeout for reliability - Graceful error handling

Example Configuration:

scout = manager.create_scout(
    name="Blog Monitor",
    type="meta",
    config={
        "tools": ["http_request"],
        "orchestration_prompt": "Monitor tech blogs for interesting articles"
    }
)

See: HTTP Request Tool Documentation

---

2. Google Search Tool 🔍

Purpose: Perform real-time web searches using Gemini Grounding.

Best for: - Finding recent news and updates - Discovering trending topics - General web research

Usage:

google_search(query="machine learning breakthroughs 2026")

Features: - Real-time search results - Automatic source attribution - Grounded responses with citations - Detailed summaries (3-4 paragraphs)

Example Configuration:

scout = manager.create_scout(
    name="News Scout",
    type="search",
    config={
        "tools": ["google_search"],
        "query": "AI regulation news"
    }
)

---

3. RSS Tool 📡

Purpose: Subscribe to and manage RSS/Atom feeds.

Best for: - Monitoring specific blogs and news sites - Following trusted content sources - Tracking updates from known publishers

Usage:

# List feeds
rss(action="list")

# Read feed
rss(action="read", feed_id="feed_123")

# Add feed
rss(action="add", url="https://blog.example.com/feed")

Features: - Database-backed feed storage - Automatic feed validation - Duplicate detection - Per-scout feed isolation

Example Configuration:

scout = manager.create_scout(
    name="Tech Blog Scout",
    type="rss",
    config={
        "tools": ["rss"],
        "feeds": ["https://techcrunch.com/feed/"]
    }
)

---

4. Reddit Tool 👾

Purpose: Fetch posts from subreddits.

Best for: - Community sentiment analysis - Finding trending discussions - Tapping into niche communities

Usage:

reddit(
    subreddit="MachineLearning",
    limit=10,
    sort="hot"  # or "new", "top", "rising"
)

Features: - Multiple sorting options (hot, new, top, rising) - Configurable post limits - Upvote and comment counts - Rate limit compliance

Example Configuration:

scout = manager.create_scout(
    name="AI Reddit Scout",
    type="reddit",
    config={
        "tools": ["reddit"],
        "subreddits": ["MachineLearning", "LocalLLaMA"],
        "reddit_sort": "hot"
    }
)

---

5. ArXiv Tool 🎓

Purpose: Search for academic papers and research.

Best for: - Finding research papers - Tracking scientific breakthroughs - Academic content curation

Usage:

arxiv_search(
    query="transformer models",
    days_back=7  # Optional: filter by recency
)

Features: - Search by keywords or ArXiv ID - Date filtering (days_back parameter) - Full paper metadata (title, authors, abstract) - Direct ArXiv links

Example Configuration:

scout = manager.create_scout(
    name="AI Papers",
    type="meta",
    config={
        "tools": ["arxiv"],
        "query": "large language models",
        "date_filter": "week"
    }
)

---

6. Browser Tool (Experimental) 🌐

Purpose: Navigate web pages with JavaScript support.

Best for: - Complex web interactions - JavaScript-heavy sites - Multi-step navigation

Usage:

browser(
    url="https://example.com",
    action="navigate"
)

Features: - Full browser automation - JavaScript execution - Multi-step interactions - Screenshot capabilities

Limitations: - Slower than http_request - More resource-intensive - Experimental stability

Example Configuration:

scout = manager.create_scout(
    name="Complex Site Monitor",
    type="meta",
    config={
        "tools": ["browser"],
        "orchestration_prompt": "Navigate to site and extract data"
    }
)

---

Tool Comparison

HTTP Request vs Browser

Feature	http_request	browser
Speed	⚡ Fast	🐌 Slower
JavaScript	❌ No	✅ Yes
CSS Selectors	✅ Yes	✅ Yes
Stability	✅ Stable	⚠️ Experimental
Resource Use	⬇️ Low	⬆️ High
Best For	Static content	Dynamic sites

Recommendation: Use http_request for most web scraping tasks. Only use browser when you specifically need JavaScript execution.

---

Combining Tools

The power of InfluencerPy comes from combining multiple tools in a Meta-Scout:

Example: Complete Research Workflow

scout = manager.create_scout(
    name="Complete Research Agent",
    type="meta",
    config={
        "tools": ["google_search", "http_request", "arxiv"],
        "orchestration_prompt": """
            1. Use google_search to find trending AI topics
            2. Use http_request to read full articles
            3. Use arxiv to find related research papers
            4. Synthesize findings into a comprehensive post
        """
    }
)

Example: News + Community Sentiment

scout = manager.create_scout(
    name="News with Sentiment",
    type="meta",
    config={
        "tools": ["rss", "reddit", "http_request"],
        "orchestration_prompt": """
            1. Check RSS feeds for latest tech news
            2. Use http_request to read full articles
            3. Check Reddit for community reactions
            4. Create a post combining news and sentiment
        """
    }
)

---

Adding Tools to Scouts

When creating a scout, specify tools in the tools array:

from influencerpy.core.scouts import ScoutManager

manager = ScoutManager()

scout = manager.create_scout(
    name="My Scout",
    type="meta",
    config={
        "tools": [
            "http_request",    # Web scraping
            "google_search",   # Search
            "arxiv",          # Research papers
            "reddit"          # Community discussions
        ],
        "orchestration_prompt": "Your orchestration instructions here"
    }
)

---

Best Practices

1. Start Simple: Begin with one tool and add more as needed
2. Choose the Right Tool: Use the comparison tables above
3. Combine Strategically: Think about the workflow (search → fetch → analyze)
4. Handle Errors: Tools return error information in responses
5. Respect Rate Limits: Don't hammer endpoints repeatedly
6. Use Selectors: When scraping, use CSS selectors for cleaner content

---

Tool Return Formats

All tools return structured data that the AI can understand:

http_request

{
    "url": str,
    "title": str,
    "content": str,
    "links": [{"text": str, "url": str}],  # Optional
    "error": str  # If failed
}

google_search

str  # Formatted text with sources

arxiv_search

str  # Paper details formatted as text

reddit

str  # Posts formatted as text with metadata

---

Next Steps

• Try the Demo: Run python examples/http_tool_demo.py to see the HTTP tool in action
• Create a Scout: Use the CLI to create a scout with your desired tools
• Experiment: Try different tool combinations to find what works best

For detailed API documentation of each tool, see the individual tool documentation files in the docs/tools/ directory.