The Memory Gap Audit is the first prompt in the Operator Second Brain Starter Pack. It surfaces the three most valuable things you know that are not written down anywhere your business can use, and ranks them by how much rebuild work they save once captured.
Run this prompt once, before you set up your operator second brain. Re-run it quarterly, or any time you finish a client engagement that exposed a pattern you had not named before.
Works best when you can reference three to five specific client engagements, projects, or decisions from the last twelve months. Have dates, rough outcomes, and one-line descriptions ready. The more concrete the input, the more useful the audit.
You are helping me audit my operator second brain gaps.
I am going to describe three to five client engagements or projects from the last twelve months. For each one, I will give you the situation, what I decided, and the outcome.
Your job is to:
1. Identify the pieces of judgment, methodology, or pattern recognition I used that are not written down anywhere.
2. Flag any that showed up in more than one engagement. Those are the highest-value gaps.
3. Rank them by how much rebuild work I save the next time I face a similar situation, if I capture them now.
4. Return the top three gaps as a prioritized list. For each one: a one-sentence name, a one-paragraph description, and a one-line hypothesis about what file it belongs in (Goals and Priorities, Session Log, Function Architecture, Brand Voice, or Working Memory).
Be direct. Do not flatter the engagements. The goal is to expose the gaps, not celebrate the work.
Here are the engagements:
[paste your engagement descriptions]
Paste the prompt into Claude or ChatGPT. Replace the bracketed section with three to five short engagement descriptions. Read the ranked output. Pick the top gap and write the first real entry against it inside the corresponding file in your Operator Second Brain workspace. Do this before you write anything else.
A strong output names three gaps in language you recognize, maps each to a specific file in your memory base, and forces you to see at least one pattern you had not noticed before. If the output feels generic, your input was too vague. Re-run with more specific engagements.
Feeding the prompt abstract descriptions instead of real engagements. Accepting the first output instead of re-running with better inputs. Filing the gaps in "someday" notes instead of capturing them inside the actual eight-file base.
The gaps that save you the most rebuild work are almost never the ones you thought were important. They are the ones you learned by doing the same thing wrong three times in a row before you corrected it. Those are the patterns AI cannot produce for you. They are the only ones worth capturing.