The Prompt Lab
Prompts that actually work, and the ones that don't.
A working bench for prompts. We publish the ones that ship, audit the ones that flop, and show the rewrites in between.
What we cover
- brief prompts
- voice-matching
- structured output
- prompt chaining
- prompt audits
- before/after rewrites
Latest in Prompt Lab
Structured Output That Survives a Model Swap: The JSON Scaffold I Actually Ship
Multi-model routing is normal now. Here is the schema-first prompt pattern that keeps objects valid when you change the brain underneath.
The 12-Prompt Eval I Run Before I Trust Any Model Upgrade
A model wave is not a migration plan. Here is the fixed task set I score before I change a default, with a pass/fail rubric you can steal.
Upgrade Your Prompt Stack for Sonnet 5 and GPT-5.6 Without Rewriting Everything
New models land and teams thrash their prompts from zero. Here is the lab method: keep the scaffold, re-test three failure modes, ship the diffs.
The De-Slop Prompt Stack: Six Prompts That Stop Claude and ChatGPT From Sounding Like a Robot
Readers now punish writing that feels machine-made. These six prompts remove the tells, and they're built to be saved once and used forever.
The Constraint Goes First
Everyone writes prompts top-down: role, context, task, then a stack of rules at the end. That's the bug.