uxBrad
@uxbrad.com did:plc:ztpxcireacf5q3w6rwsaz7rm
I'm Brad Cooper a UX Designer. This is my secondary account which is much more real-time than my federated one: https://bsky.app/profile/uxbrad.com.web.brid.gy. Follow me wherever you follow people.
Bluesky’a katılma tarihi:
- Takipçi
- 110
- Takip edilen
- 262
- Post
- 271
- Sıralamaya girdiği gün
- 0
Özet
uxBrad (@uxbrad.com), 26 Kasım 2024 tarihinde katılan bir Bluesky kullanıcısı. 110 takipçisi var, 262 hesabı takip ediyor ve 271 post paylaştı.
Sıralama geçmişi
Bu hesabın postlarının Debya günlük listesine girdiği en iyi günler.
Bu hesap henüz günlük sıralamaya girmedi.
Son beğendiği postlar
Bluesky ağındaki herkese açık beğeniler.
And setting aside AI questions, beauty is contextual, depends upon other work. Beethoven wrote beautiful music but new music like that today would be derivative and cheesy. Similarly, an app or doc may be beautiful the first time ... yet tedious in its 100th clone. (cf. LLM presentation templates)
Postu gösterFor the HTML→WCAG→fix loop, I’d run the same page through two fix/rescan cycles. PASS = the original issue disappears, no new violation is introduced, and the agent doesn’t oscillate between two “fixes.” If useful, I can turn that into a small independent test.
Postu gösterAgreed. Keeping that written brief in a reusable file and adding one line each time the output misses makes it better week by week.
Postu gösterYes. The minimum viable prompt: what changes, what it touches, and how to undo — all before the button. Bonus: scope it to this task, not the tool forever. A yes/no you can't evaluate is a ritual, not consent.
Postu gösterThe design problem of agent scope is real. We have a connector that helps solve the "black box" issue by enforcing engineering rigor through deterministic validation rather than just text generation: https://vinkius.com/en/ai-agent-connect/code-integrity-prover
Postu göster
Blokladığı hesaplar
Bluesky’da bloklar herkese açık kayıtlardır.