PixelHush: auto-hide secrets in code editors and browsers during screen recording. From 3 wasted hours to shipped product in 2 weeks
PixelHush: auto-hide secrets in code editors and browsers during screen recording. From 3 wasted hours to shipped product in 2 weeks.
The trigger: I recorded a 3-hour video about building a Telegram agent. I discovered that about a third of it had API keys and tokens visible on screen — keys that can't be easily reissued. The video's length made manual blurring impractical, so it's still sitting on my drive, unedited. Looked for a tool that auto-hides secrets when recording — nothing existed that worked automatically.
What I built: PixelHush — a macOS menu bar app + VS Code extension (pixelhush.dev). The app detects recording software (macOS Built-in, OBS Studio, Zoom, Chrome, and more). The extension masks secret values visually using decorators. Files are never modified.
Stack:
macOS agent: Swift 5.9, AppKit, CGWindowList API — 312 KB binary
VS Code extension: TypeScript, fork of zeybek/camouflage (MIT) — 216 KB VSIX
Bridge: flag file at ~/.pixelhush/recording (no server, no WebSocket)
License verification: Polar.sh with 7-day offline grace
Timeline:
Day 1–2: Forked camouflage, built the recording watcher
Day 3–5: macOS agent with bundle ID detection
Day 6–8: Window layer inspection for Zoom/Discord/browsers
Day 9–10: Free/Pro gating, license system
Day 11–12: Onboarding flow, DMG packaging
Day 13–14: Polish, 312 tests total
Pricing:
Free: auto-detection of OBS Studio, QuickTime, Zoom, Screenshot (Cmd+Shift+5), Chrome. Manual toggle (Cmd+Shift+H). 48 patterns, all file types — works standalone without the macOS app too
Pro: $49/yr Early Bird — adds Loom, Discord, Teams, Slack, CleanShot X, Screen Studio, ScreenFlow, Kap. Click-to-copy hidden values. Custom key patterns. Chrome Extension for browser protection
Decisions I'd make again:
File-based bridge instead of WebSocket. Zero complexity, survives restarts.
Forking an existing MIT extension instead of building from scratch. Saved weeks.
Free tier that's genuinely useful on its own. Covers the 4 most common recording apps without paying anything.
Decisions I'm still questioning:
macOS only for now. The extension works anywhere, but auto-detection requires OS-level APIs. Windows would need a different approach entirely.
$49/yr Early Bird. Feels underpriced — but I'd rather have early users with real feedback than optimize pricing before product-market fit.
What's shipped / what's coming:
✅ macOS app + VS Code extension — available at pixelhush.dev
✅ Chrome Extension — syncs with the macOS agent for browser protection
✅ VS Code Marketplace listing
🔜 Hide secrets in .py, .ts, .js files
Would love feedback on the pricing or the approach. Has anyone else dealt with the "secrets on camera" problem?
Love the "scratch your own itch" origin story here — those are always the most compelling products because you deeply understand the pain. The 2-week timeline is impressive too, and forking an MIT extension instead of building from scratch is such a smart move that more founders should adopt. We took a similar approach when building our ad creative tool — reusing open-source rendering libraries instead of writing a custom canvas engine saved us weeks and let us focus on the actual UX problem (making ad creation feel instant rather than like a design project). On the pricing question: $49/yr feels reasonable for early bird, and I think you're right to prioritize feedback over revenue at this stage. One thing I'd watch is whether your free tier is almost too generous — if the 4 most common recording apps are covered free, what's the conversion trigger? Might be worth tracking which Pro features actually drive upgrades once you have more data.