security report automator

The problem

A first-pass external recon report is repetitive work: enumerate subdomains, check exposed services, correlate CVEs, review SSL/TLS, then write it all up. I wanted that turned into a single input — one domain — that returns a clean, executive-ready report without any active scanning or manual write-up.

The approach

A domain and delivery email kick off a passive-only pipeline. Recon sources are queried in parallel, the findings are scored and summarized by an LLM, rendered into a branded PDF, and delivered to the inbox — with a Slack ping on completion. Nothing touches the target beyond public data.

How it works

1. 01
   initializing

   A submitted domain + email spins up the reconnaissance pipeline and issues a scan ID.

2. 02
   reconnaissance

   crt.sh, HackerTarget, Shodan, SSL Labs, and NVD are queried in parallel — passive sources only, no active probing.

3. 03
   ai analysis

   Groq (Llama) scores the attack surface out of 100 and drafts an executive summary, findings, and remediation steps.

4. 04
   pdf rendering

   Results are rendered into a branded, multi-page assessment report with a cover, risk score, and detailed findings.

5. 05
   complete

   The PDF is emailed to the requester, a Slack notification fires, and the run is logged to Supabase.

Notes

• Passive-only by design — safe to point at any domain without authorization concerns.
• Risk score is derived from exposed services, known CVEs, SSL/TLS posture, and subdomain exposure.
• Runs on free-tier infrastructure end to end (n8n, Groq, Supabase).

Walkthrough