ObscuronLabs™ — AI Automation Suite
Multi-agent email pipeline · B2B lead scraper · n8n Discord automation
ObscuronLabs™ builds AI agents, automation systems, and digital intelligence infrastructure for businesses. This repository contains the open-source automation toolkit that powers our backend pipelines.
Website: obscuronlabs.com
What's in this repository
| File | Description |
|---|
core/agents.py | Three-agent AI pipeline (triage → strategy → response) |
core/scraper.py | B2B lead scraper — extracts emails, phones, and company names from public websites |
main.py | CLI entry point |
workflows/obscuron_automation.json | n8n workflow — imports directly into your n8n instance |
requirements.txt | Python dependencies, pinned versions |
.env.example | Configuration template |
Quick Start
1. Install dependencies
python -m venv venv
source venv/bin/activate # Windows: venv\Scripts\activate
pip install -r requirements.txt
2. Configure environment
cp .env.example .env
# Add your API keys to .env
Required (choose one):
3. Run the agent pipeline
python main.py pipeline \
--email "[email protected]" \
--message "We need help automating our weekly reports."
Output: triage analysis, delivery strategy, and a draft email response.
4. Scrape leads
# Scrape specific websites
python main.py scrape --urls stripe.com hubspot.com
# Search and scrape (requires SERPAPI_KEY)
python main.py search --query "logistics companies Berlin" --results 20
Output: timestamped CSV in the outputs/ folder.
Agent Pipeline
[Input Email] → Maya (triage) → Leo (strategy) → Sophia (response draft)
Each agent uses a different model:
- Maya —
gpt-4o-mini (triage)
- Leo and Sophia —
gpt-4o (strategy and client-facing output)
Results saved to outputs/<timestamp>_<email>.json.
Lead Scraper
Per website, the scraper:
- Fetches the homepage
- Extracts emails and phones
- Auto-discovers contact/about pages if no email found
- Saves everything to CSV
Works on any public business website. Respects configurable request delays.
With SerpAPI: Search Google and auto-scrape the top N results.
python main.py search --query "SaaS companies London" --results 15 --delay 2.0
n8n Workflow Setup
- Open your n8n instance
- Click Import from file → select
workflows/obscuron_automation.json
- In each HTTP Request node, replace
YOUR_OPENROUTER_API_KEY with your key
- Update Discord nodes with your Guild ID and Channel IDs
- Activate the workflow
The workflow posts AI-generated reports to Discord channels on a schedule.
Configuration Reference
| Variable | Required | Description |
|---|
OPENAI_API_KEY | If not using OpenRouter | OpenAI API key |
OPENROUTER_API_KEY | If not using OpenAI | OpenRouter API key |
USE_OPENROUTER | No (default: false) | Set true to route through OpenRouter |
SERPAPI_KEY | For search-based scraping | Free tier at serpapi.com |
DISCORD_BOT_TOKEN | For Discord notifications | Discord bot token |
OUTPUT_DIR | No (default: outputs) | Output directory for results |
Extending the Toolkit
Add an agent: Subclass BaseAgent in core/agents.py, define a SYSTEM prompt, add it to ObscuronPipeline.process().
Add a scraping source: Add a method to ObscuronScraper in core/scraper.py. CSV output format stays the same.
Add an n8n trigger: Import the workflow, add a Webhook node, connect it to the existing HTTP Request chain.
License
See LICENSE.txt.
Contact
For custom AI solutions, automation engagements, or questions about this toolkit, reach out via obscuronlabs.com.
Built by ObscuronLabs™.
