I’ve always felt that reputation management tools have a major blind spot: they are great at tracking web and social media text, but they completely overlook what’s being actually spoken inside YouTube videos.
To solve this for myself at my previous job, I built a hacky Excel VBA script. It got the job done, but it was clunky, tied to Excel, and I kept forgetting to run it. I really wanted to build a proper, lightweight standalone version.
There was just one problem: I wanted to build it in Go, and my Go experience was exactly zero.
The AI Advantage: Architecting vs. Typing
Instead of spending weeks learning syntax, I tried a different approach. I brought the logic, system design, and architecture from my VBA days, but I essentially used AI as my hands to write the actual Go code. I designed the core logic, and directed the AI to handle the syntax and implementation.
The result is SentryScript, a portable Windows app that monitors specific YouTube channels for your keywords. When a mention is found, it provides a clickable, timestamped URL leading exactly to the moment the word was spoken.
Key Features of SentryScript:
Privacy-First & Local: No cloud involved. Subtitle parsing and regex matching happen entirely in your RAM. It never writes subtitle content to your disk or sends it to a remote server. Only the final metadata (date, title, URL) is saved locally.
Multilingual: It automatically translates your target keywords into 11 languages to scan subtitles, even in foreign language videos.
Lightweight: yt-dlp as a backend, it fetches metadata.
Using the scheduling feature, these tasks are repeated at intervals of 12, 24, or 48 hours.
Portable: No installation required, and absolutely no telemetry.
A Quick Note on the "Unknown Publisher" Warning
As an independent developer, I haven't invested in expensive code-signing certificates just yet. Because of this, Windows SmartScreen will likely flag the EXE. For full transparency, I’ve included a VirusTotal scan right in the repository.
You can check out the GitHub repo here: [https://github.com/SentryScript/SentryScript]
How does the AI-generated code hold up? Let me know!
Top comments (0)