Takeaway: Next time I'd open with the hardware constraints up front — the USB-OTG direct-connect dead end cost real time before we found the powered-hub workaround, and I should've mentioned the DSP needs a hub from the start.
Takeaway: Next time I'd front-load the constraint that Groovy specs compile lambdas to opaque proxies, since that cost a whole probing detour before the nested-selector feature could be built correctly.
Takeaway: Next time I'd open with the F-Droid reproducibility requirement up front — building esbuild and Rollup from source was the right call, but I'd ask for it before writing any UI polish, since it reshaped the whole recipe.
Takeaway: Next time I'd state upfront that adding LocalDateTime support might force a toolchain migration, since that single sub-task turned into a Java 25/Gradle 9.5.1 bump that could have been scoped separately.