The four heights
The same task, four distances: today's deadline, the next reviewer, the stuck moment, the pattern.
Execute — do the immediate task
+I need a random token for a password-reset link for user 16742. Generate a cryptographically secure…
Execute — do the immediate task
+I need a random token for a password-reset link for user 16742. Generate a cryptographically secure random string of 32 URL-safe characters, base64-url encoded, and attach it to resetTokens with an expiry 15 minutes from now. Return the token so I can send it in the email.
Pasted it? When the reply comes back, push once: ask it to sharpen the weakest part. — Did this prompt help?
Improve — make it easier to accept
+Before I put random tokens into production, make the generation easy to audit: show the entropy…
Improve — make it easier to accept
+Before I put random tokens into production, make the generation easy to audit: show the entropy bits, write a one-line comment explaining why the method is cryptographically secure, and flag any uses of Math.random or non-cryptographic APIs. Also provide a deterministic fallback for tests that can be toggled without changing the production path.
Pasted it? When the reply comes back, push once: ask it to sharpen the weakest part. — Did this prompt help?
Decide — diagnose the stuck moment
+I just generated a password-reset token for user 16742 and a support case says the token matched…
Decide — diagnose the stuck moment
+I issued a reset token and a user reported they received the same token twice.
I just generated a password-reset token for user 16742 and a support case says the token matched another recent token. I suspect a non-cryptographic PRNG slipped into the code path, or I accidentally seeded a deterministic routine for tests. I can’t tell whether the bug is in token creation, storage, or a test harness. What is the most likely cause and the immediate code checks I should run to confirm and fix it now?
Pasted it? When the reply comes back, push once: ask it to sharpen the weakest part. — Did this prompt help?
Become — change the pattern
+Over time I see intermittent token collisions or predictable tokens in logs and it costs hours in…
Become — change the pattern
+Tokens occasionally collide in logs
Over time I see intermittent token collisions or predictable tokens in logs and it costs hours in incident triage. I keep using ad-hoc token helpers that are convenient for dev but unsafe for prod. What practice and minimal tooling change will stop these incidents, and how should I handle testability without weakening security in production?
Pasted it? When the reply comes back, push once: ask it to sharpen the weakest part. — Did this prompt help?
Next to this one
Other programming language work people do in JavaScript.
Every task here came from the work, not from a feature list — which is why the prompts name what you want done and never the button that does it. The tool changes; the work does not.
Copyright © LLOS.ai · 2026 — original pedagogy, voice, and design — all rights reserved.
The rest of the map
Same library, five ways in.