Help me set a time budget and stop conditions for this AI task before I start, so I don't burn three hours and a hundred dollars on something that wasn't going to work.
The task:
{{describe the task}}
The tool / model: {{Cursor, Claude Code, ChatGPT, Aider, etc.}}
Spec clarity (1-5): {{1 = vague feeling, 5 = exact requirements with examples}}
Domain familiarity (1-5): {{1 = never touched this, 5 = expert}}
Decide for me:
1. **Time budget** — minutes before I should pause and reassess (not "give up" — pause)
2. **Cost budget** — rough $ ceiling at the chosen tool
3. **Stop conditions** — concrete signals that mean "stop iterating, do something else":
- Loop signal (e.g., "same fix tried twice")
- Quality signal (e.g., "diff is bigger than expected for this scope")
- Confidence signal (e.g., "AI starts hedging instead of answering")
4. **Reassess question** — when I hit the time budget, what should I ask myself? ("Is the prompt wrong, the tool wrong, or is the problem under-specified?")
5. **Fallback move** — if I have to give up: smaller scope? Different tool? Write it myself? Ask a human?
Calibrate budgets to the scores. Vague spec + unfamiliar area = shorter budget, not longer.budgetingworkflowai-discipline