Skip to content

Getting Started

Prerequisites

Before you begin, ensure you have the following:

  1. Python 3.9 or higher: Check with python --version.
  2. Telegram Bot (Required): Used to receive scout reports and review drafts.
    • Message @BotFather to create a bot and get your Bot Token
    • Message @userinfobot to get your Chat ID
  3. API Keys: You will need API keys depending on your use case:
    • Google Gemini API Key (Required for AI features): Get it from Google AI Studio.
    • X (Twitter) API Keys (Optional, only for posting): You need a Developer Account with "Read and Write" permissions. Get them from the X Developer Portal.
    • Substack Cookies (Optional, only for paywalled content): See the Substack Setup Guide for details.

Installation

Installation

InfluencerPy is currently available via source installation.

  1. Clone the repository: 

    git clone https://github.com/cristianleoo/InfluencerPy.git
cd InfluencerPy

  2. Install dependencies: 

    pip install -e .

Initial Configuration

InfluencerPy uses a configuration wizard to make setup easy. You don't need to manually edit config files unless you want to.

1. Run the CLI

Start the application:

influencerpy

2. Setup Credentials

On your first run, you will see the Credential Setup Guide.

  1. The wizard will ask for your Gemini API Key (required for AI features).
  2. It will ask for your Telegram Bot Token and Chat ID (required for receiving reports).
  3. Optionally, add X API credentials if you plan to post to Twitter.
  4. These are saved securely in a .env file in your project root.

3. Configure AI Settings

By default, InfluencerPy uses Gemini Flash for fast processing. You can customize this:

  1. Select Configure AI Settings from the main menu.
  2. Choose your provider: Gemini or Anthropic.
  3. Set a specific Model ID (e.g., gemini-2.5-flash for speed, or claude-sonnet-4 for quality).
  4. Adjust the Temperature (default 0.7). Higher values make output more creative; lower values make it more factual.

4. Optional: Telemetry (Langfuse)

To trace and debug your Scout's AI reasoning, you can enable Langfuse integration.

  1. Select Configure AI Settings -> Langfuse (Tracing) from the menu.
  2. Enter your Host, Public Key, and Secret Key.
  3. This setting is global: once enabled, all Scouts will report traces to your Langfuse project.

Creating Your First Scout

Let's create a content discovery scout:

  1. Run influencerpy and select "Scouts" → "Create Scout"
  2. Choose Intent: Select "🔍 Content Discovery" (this finds and lists content)
  3. Scout Type: Choose a source (RSS, Reddit, Arxiv, etc.)
  4. Configure Source: Enter feed URLs, subreddits, or search queries
  5. Tip: For RSS feeds, you can add multiple feeds separated by commas (e.g., https://feed1.com/rss, https://feed2.com/rss)
  6. Schedule: Set when it should run (Daily, Weekly, or Manual)
  7. Done!: Your scout is ready

Running Your Scout

Manual Run: - From CLI: Select the scout and choose "Run Scout" - From Telegram: Send /scouts and click "🚀 Run Scout"

Scheduled Run: - The bot will automatically run scouts based on their schedule - Reports are sent to Telegram

What You'll Receive

For Scouting Intent scouts, you'll receive: - A curated list of content items - Summary of each item - Links to original sources - Delivered via Telegram for easy reading

For Generation Intent scouts (optional), you'll receive: - Draft social media posts - Ready to post to X (Twitter) or copy/paste elsewhere

Example: Multi-Feed AI Research Scout

Here's a complete example of creating a content discovery scout that monitors multiple AI research sources:

Configuration

Scout Name: AI Research Digest
Intent: 🔍 Content Discovery (Scouting)
Scout Type: 📡 RSS
RSS Feed URLs: https://tldr.takara.ai/api/papers, https://bair.berkeley.edu/blog/feed.xml, https://research.google/blog/rss/, https://news.mit.edu/rss/topic/artificial-intelligence2, https://news.microsoft.com/source/topics/ai/feed/
Schedule: Daily at 9:00 AM

What This Does

This scout will:

  1. Monitor 5 AI research sources simultaneously:
  2. Takara AI TLDR Papers
  3. Berkeley AI Research (BAIR) Blog
  4. Google Research Blog
  5. MIT AI News

  6. Microsoft AI News

  7. Explore ALL feeds comprehensively: The AI automatically reads entries from ALL 5 feeds (not just one), gathering diverse content across all sources

  8. Analyze and select the best content: The AI reviews articles from all feeds and selects the most relevant and interesting ones

  9. Deliver a curated report to Telegram with:

  10. Article titles and summaries
  11. Key insights from each piece
  12. Links to original sources
  13. Source attribution (showing which feed each article came from)
  14. Delivered every morning at 9 AM

Sample Output

# 📚 AI Research Digest - Content Discovery
*Found 6 interesting items from 5 sources*

## 1. New Advances in Multimodal Learning
Summary: Researchers at Berkeley demonstrate significant improvements in vision-language models...
🔗 Source: https://bair.berkeley.edu/blog/2025/...

## 2. Scaling Language Models: Latest Insights
Summary: Google Research shares findings on training efficiency...
🔗 Source: https://research.google/blog/...

[... more items ...]

Why Multiple Feeds?

  • Broader coverage: Get diverse perspectives from academia and industry
  • All feeds explored: The agent automatically reads from ALL feeds, not just one
  • Better selection: More content from multiple sources means the AI can pick truly standout articles
  • Single digest: One consolidated report instead of multiple notifications
  • Source attribution: See which feed each article came from
  • Efficient: Set up once, runs automatically every day

Next Steps