🔁 Core Workflow Flow
[1] Describe what you want
[2] Approve/refine AI suggestion
[3] Confirm canvas updates
[4] Save/track decisions
[5] Repeat in modular steps
Your Role (What You Handle)
1. Start with Intent
- Always give clear goal or desired outcome, not just “write code.”
- Ex: “I want a simple REST API to track books” or “Let’s plan a React Native screen for hunt details.”
2. Approve Canvas Target
- AI will propose what canvas it wants to update.
- You must confirm it’s the right one before changes are made.
- Be explicit: “Yes, update the
hunt_editor
canvas.”
3. Log Key Decisions (or ask AI to write it for you)
- If a name, library, structure, or behavior is chosen, log the why.
- Use a
Decisions.md
,README.md
, or whatever doc format fits.
4. Keep It Modular
- Ask for one feature, function, or module at a time.
- Avoid requesting giant full-app dumps unless you’re just prototyping.
5. Request Summaries
-
After any major change or code output:
- Ask: “Give me a summary of what you just did”
- Ex: files affected, rationale, what to test
Good Prompts to Keep Handy
- “Start with a plan in markdown.”
- “Write a
Decisions.md
entry for that choice.” - “Let’s do this one module at a time.”
- “What do we need to build this feature?”
- “Summarize changes in bullet points.”
- “What canvas are you planning to update?”
Caution Zones
Danger Zone | Your Defense |
---|---|
AI changes wrong canvas | Approve target canvas before update |
No memory of why something was done | Ask for or write a Decisions.md note |
AI gives giant blob of code | Ask to break it up, one part at a time |
You forget context later | Save markdown planning and summaries |
Feature feels wrong after build | Trust your gut — request a rewrite |