Poirot Surpassed 100 GitHub Stars
Poirot just surpassed 100 stars on GitHub.
I built it in a single weekend while on parental leave. The idea was simple: I was using Claude Code daily, but reading raw JSONL session transcripts to understand what the AI actually did was painful. So I built a native macOS app to investigate those sessions visually.
The “meta” part? I built the app using the very AI it was designed to complement.
Swift 6, strict concurrency, SwiftUI, the new Swift Testing framework. Less than 6 MB. 100% offline. No login, no tracking, no analytics. Open source under MIT.
What started as a personal scratch-your-own-itch tool has resonated with the developer community far more than I expected. The feedback and adoption has been incredible. Feature requests, contributions, and messages from developers who use it daily to review what their agents are doing.
The lesson I keep relearning: build something you genuinely need, ship it fast, open source it, and let the community decide if it matters.
Install via Homebrew: brew install --cask poirot