WTDN - What To Do Next
Input: $ARGUMENTS
Core Principles
-
Simple means simple. This skill has no routing, no classification, no sub-orchestration. It asks a few questions and gives one answer. If the user needs complexity, they’ll find it elsewhere.
-
One move, one reason. The output is a single action and a single reason. Not three options. Not a comparison. One thing to do and why that one.
-
Bias toward action. When two next moves seem equally valuable, pick the one that starts sooner and produces a visible result faster.
-
“I don’t know” is fine. If the user can’t articulate where they are, that’s data. The next step is to figure out where they are.
Steps
Step 1: Where Are You Right Now?
State the current situation in one sentence based on $ARGUMENTS.
SITUATION: [what's happening right now — project state, problem state, or "I don't know"]
If the user hasn’t provided enough context, ask: “Where are you right now with this? What have you done so far?”
Step 2: What Are You Trying to Reach?
GOAL: [the end state or outcome they want]
If unclear from input, use: “What does ‘done’ look like?”
Step 3: What’s the Gap?
GAP: [the single biggest thing between SITUATION and GOAL]
Step 4: What’s the Single Most Valuable Next Move?
NEXT MOVE: [one specific action]
WHY THIS ONE: [one sentence — why this move over any other]
Pick the move that:
- Closes the biggest part of the gap
- Can be started immediately
- Produces a result you can see or measure
Step 5: Do It
DO THIS NOW: [restate the action as a direct instruction]
TIME ESTIMATE: [how long this should take]
YOU'LL KNOW IT WORKED WHEN: [observable result]
Quick Mode
If the user adds “quick” or wants speed, skip to three questions:
QUICK MODE
==========
1. Where are you? [one sentence from input]
2. What's blocking progress? [one sentence]
3. DO THIS: [one action that addresses the block]
No analysis. No explanation. Just the answer.
Output Format
WHAT TO DO NEXT
===============
SITUATION: [where you are]
GOAL: [where you're headed]
GAP: [what's between]
→ NEXT MOVE: [the one thing to do]
WHY: [one reason]
TIME: [estimate]
DONE WHEN: [observable result]
Failure Modes
| Failure | Signal | Fix |
|---|---|---|
| Gave multiple options | Output contains “or” or numbered alternatives | Pick one. That’s the whole point |
| Too vague | ”Work on the project” | What specific action, in what tool/place, producing what? |
| Overthinking | More than 2 minutes on this skill | Use Quick Mode instead |
| User doesn’t know their situation | Can’t answer Step 1 | Next move is: “Spend 5 minutes writing down what you know and don’t know” |
Integration
- Simpler than:
/next(which classifies intent and routes to other skills) - Simpler than:
/nowwt(which handles aftermath situations) - If user needs deeper analysis: route to
/next - If user is stuck, not just “what’s next”: route to
/unstk - If user just needs the absolute minimum: route to
/nstep