Vol. 1 gave you three codes and a promise. This is the promise kept: the full ten-code pack. Same rules as before - paste the code before your request, stack max two per conversation, re-paste if the model drifts.
The originals · 01–03
Quick recap from Vol. 1 - full prompts there.
- CODE 01 · THE OPERATOR - senior operator mode: one recommendation, reasons after, no caveats.
- CODE 02 · THE INTERROGATOR - asks the 3 questions that most change the output before doing anything.
- CODE 03 · THE RED TEAM - attacks your plan: 3 weakest assumptions, ranked by damage, cheapest test for each.
New codes · 04–06
Code 04 · The compressor
Answer in 1/5 the length you normally would. Every sentence must earn its place. If I want depth I'll ask.
Use when answers keep ballooning and you're skimming instead of reading.
Code 05 · The sparring partner
Before you agree with anything I say, argue the strongest case against it. Then tell me which side you actually find more convincing and why.
Use when you catch the model agreeing with everything. Kills the yes-man.
Code 06 · The ship-it filter
We ship in 30 minutes. Given that constraint, what do we cut, what do we keep, and what's the single highest-leverage improvement we make first?
Use when a project is stuck in polish loops. Deadlines beat perfectionism.
New codes · 07–10
Code 07 · The explainer
Explain this back to me like I'm a smart 12-year-old, then list the 3 things you'd need to check before betting money you understood it right.
Use on anything you're about to act on but only half understand.
Code 08 · The delegator
Turn this goal into instructions so specific that a brand-new assistant with zero context could execute them without asking me anything.
Use before handing work to an agent, a VA, or future-you. Vague goals die in handoff.
Code 09 · The post-mortem
That didn't work. Write the 5-line post-mortem: what we expected, what happened, the most likely cause, the test that confirms it, the fix.
Use the moment something fails - before you try again blind.
Code 10 · The closer
Before we wrap: what did you learn about how I work that you'd carry into our next session? Write it as 3 rules a future assistant should follow.
Use at the end of every serious session. The rules compound - save them.
Run the stack
| Situation | Code |
|---|---|
| Output too long, skimming everything | 04 · Compressor |
| Model agrees with everything you say | 05 · Sparring Partner |
| Stuck polishing instead of shipping | 06 · Ship-It Filter |
| About to act on something half-understood | 07 · Explainer |
| Handing work to an agent or assistant | 08 · Delegator |
| Something just failed | 09 · Post-Mortem |
| Wrapping any serious session | 10 · Closer |
These work in any AI chat - Claude, ChatGPT, anything that reads instructions. Vol. 3 is coming: follow opusjake.ai so you don't miss it.