only what was needed

selfactor
deploy
- fixed the .plan site deploy after the server package crossed Lambda's unzipped size limit
- the tracing root was the repository root, so source artwork, blog media, deployment state and old artifact zips were packed into the function
- public media is served from the bucket as static assets again, and only the plan MDX the pages read is re-included
- server bundle 424 MB to 44 MB
artifacts
- built the Furious Torpedo Zerk shell as a presentation layer over the audited artifact baseline
- the layer moves existing DOM rather than cloning listeners, and the build prints sizes and hashes and fails unless stripping the two marked blocks reproduces the baseline byte for byte
- opened it as a Selfactor artifact with a PRD revision for the next pass
lab
- traced the Lab app relaunching the moment it was quit to two launchd jobs
- the restore job had been resubmitted 9,916 times and reran the canonical launch on every exit
- read both scripts before unloading either; both jobs are gone and the app stays off
materials
- completed the NVIDIA Inception update deck
hal
- published the previous entry and confirmed the dated URL served