55 lines
1.2 KiB
Markdown
55 lines
1.2 KiB
Markdown
```
|
||
Read TO-DOS.md in full.
|
||
Summarize:
|
||
- Goal
|
||
- Current Phase
|
||
- Active Tasks
|
||
Do not write code until this summary is complete.
|
||
```
|
||
|
||
```
|
||
/superpowers:brainstorm todo X <-- will create a design doc in the docs/plan
|
||
```
|
||
|
||
```
|
||
/superpowers:write-plan for the design created
|
||
```
|
||
|
||
```
|
||
/superpowers:subagent-driven-development
|
||
```
|
||
|
||
```
|
||
read docs/TEST_PLAN.md in full
|
||
|
||
Summarize:
|
||
- Goal
|
||
- Current Phase
|
||
- Active Tasks
|
||
Do not write code until this summary is complete.
|
||
```
|
||
|
||
# Claude Code “Decision OS” — Usage Guide
|
||
|
||
This guide explains **how to actually use** your Claude Code command set in real-world product and engineering work.
|
||
It’s written for **shipping software**, not prompt theory.
|
||
|
||
If you follow this process, you will:
|
||
- Catch bad assumptions early
|
||
- Avoid rework
|
||
- Get clearer designs, safer rollouts, and better tests
|
||
- Stop Claude from jumping straight to mediocre solutions
|
||
|
||
---
|
||
|
||
## Core Mental Model (Read Once)
|
||
|
||
Every command enforces the same lifecycle:
|
||
|
||
1. **Phase 1 — Interview (Questions Only)**
|
||
Claude asks clarifying questions. No code. No designs. No solutions.
|
||
|
||
2. **START Gate**
|
||
You explicitly unlock execution by typing:
|
||
|