Help-Center Crawler

​

Flow

WEBSITE → SENSO WORKSPACE (ingest → vectorize) → TERMINAL Q&A

​

Grab the code

The fully-commented script (cli_support_hub.py) lives in our public repo: ➡️ https://github.com/AI-Template-SDK/api-quickstart/tree/main/support-hub Clone, tweak the logging, or embed the logic in your own apps.

​

Why run it?

• Merge scattered help pages into one semantic search bar.
• Watch every ingest + search step — great for debugging.
• Fork a zero-dependency demo and extend it in minutes.

​

How it works (cli_support_hub.py)

1. Scrape — Firecrawl pulls clean Markdown for each URL.
2. Upload — POST /content/raw; poll GET /content/{id} until ready.
3. Ask — CLI calls POST /search, returns answers with cited chunks

​

Endpoints touched

• POST /content/raw - create each crawled page as content • GET /content/{content_id} - poll until ingestion finishes • POST /search - semantic Q&A over everything you just loaded

​

Setup

export SENSO_KEY="sk_prod_xxx"
export FIRECRAWL_KEY="fc_live_xxx"

pip install requests rich
python cli_support_hub.py https://docs.acme.com/faq https://acme.com/manual

When you see ✓ Indexed, start asking questions directly in the terminal.

​

What you’ll learn

• The life-cycle of a raw content object • Real-world timings for chunking & vectorisation • How /search cites specific content IDs for traceability • A reusable pattern: post → poll → act