I wanted to know who my computer was talking to — so I built it (and strangers shipped the Windows port)

Description and use

The author built Snitch to provide real‑time, privacy‑focused visibility into a computer’s network connections. Community contributions finished the Windows port, and the project gained traction after a Chinese tech newsletter feature.

Problem

The author wanted a live, visual answer to which app, server, and country the computer was communicating with at any moment.

Outcome

Snitch achieved rapid community adoption, with a Windows port completed by external contributors and a 20% traffic spike from Chinese readers after a newsletter feature.

Workflow

• 1. Capture packets using ctypes→libpcap on macOS/Linux or Npcap on Windows.
• 2. Parse IPv4/IPv6/TCP/UDP/DNS/TLS‑SNI with hand‑rolled decoders (~2K lines).
• 3. Store data in a SQLite ring buffer (24h retention).
• 4. Serve data via FastAPI + WebSockets to a React + D3 frontend with a world map.
• 5. Perform local geolocation using MaxMind/DB‑IP files; no outbound calls unless user consents.

Evidence

I built Snitch. Open source, AGPL, EN/FR UI, ~400 stars in three weeks.

Source

The constraint I set myself: no heavy dependencies, no telemetry, ever.

Source

Users filed bug reports with root‑cause analysis included — not "it doesn't work", actual stack traces and the faulty capture path identified

Source

A Chinese tech newsletter (ruanyifeng.com) featured the project — suddenly ~20% of traffic was Chinese readers staring at an English README.

Source