| name | report-writer |
| description | Use this to assemble the final audit in the exact order that lands - one page, ten findings, then depth. |
Report Writer
Fixed Order
Do NOT open with an academic dump. The order is the product:
- HOW YOU ACTUALLY WORK - one page maximum. The dominant work loop in
direct concrete language.
- THE 10 THINGS YOU MAY NOT REALIZE - each with finding, why it
matters, strongest evidence, counterevidence, confidence, what to do
differently. Surprising + actionable beats flattering.
- BELIEF VS BEHAVIOR table.
- DECISION OPERATING SYSTEM - how I begin, reduce uncertainty, choose,
react to failure, verify, call something done.
- STRENGTH -> SHADOW PAIRS - evidence-backed only.
- RECURRING FAILURE LOOPS - ranked by expected impact.
- WHAT I ACTUALLY LEARN - fastest areas, gaps, transfer wins.
- DELEGATION FRONTIER - four buckets, task-specific:
DELEGATE COMPLETELY / DELEGATE + VERIFY / COLLABORATE DEEPLY /
KEEP HUMAN-LED.
- ME x MODEL MATRIX - which model for which work, from outcomes.
- WHERE AI MAKES ME BETTER / WORSE / ONLY FEELS EASIER.
The third category exists only if evidence supports it.
- HIDDEN BOTTLENECKS - never assume coding speed is the constraint.
- PERSONAL OPERATING MANUAL - about 10 rules:
WHEN [observable situation] DO [specific behavior]
BECAUSE [evidence-backed reason].
- EARLY-WARNING SIGNALS - "if X happened twice, stop and do Y."
- THREE CHANGES ONLY - behavior, evidence, expected benefit,
difficulty, exact replacement behavior, measurement.
- LEARNING ROADMAP - LEARN DEEPLY / UNDERSTAND ENOUGH TO SUPERVISE /
SAFE TO OUTSOURCE.
- WHAT I CANNOT KNOW FROM THESE DATA - unsupported conclusions,
missing data, confounders, selection bias, retrospective limits,
hypotheses needing prospective tests.
Success Criterion
The reader should repeatedly think:
"I didn't realize I did that."
then
"Now I know exactly what to do differently."
It FAILS if it mainly says: ambitious, curious, moves fast, likes
technology, detail-oriented, communicate clearly, take breaks.
Discipline
Outcome measures only. State uncertainty on cycle times. No clinical or
personality labels anywhere. Methodology and sources documented separately
from the personal report (see SOURCES.md).
One-Line Memory
One page of truth before fifty pages of proof.
