Think
Input: $ARGUMENTS
Step 1: What Kind of Thinking Is Needed?
Read the input and classify the thinking required.
INPUT SUMMARY: [One sentence restating what the user needs]
THINKING TYPE NEEDED:
| Type | Signal | Match? |
|---|---|---|
| Analytical | Break something down, understand structure | [yes/no] |
| Critical | Test a claim, find flaws, validate | [yes/no] |
| Creative | Generate options, find novel solutions | [yes/no] |
| Strategic | Plan, prioritize, sequence actions | [yes/no] |
| Empathetic | Understand perspectives, feelings, motivations | [yes/no] |
| Systemic | Understand how parts interact, feedback loops | [yes/no] |
| Evaluative | Judge quality, compare, rank | [yes/no] |
PRIMARY THINKING TYPE: [the dominant one]
SECONDARY: [if applicable]
Step 2: Select the Right Tool
Based on the thinking type, choose the best skill or framework.
FOR THIS THINKING TYPE, THE BEST APPROACH IS:
SKILL: [/skillname] — because [reason it fits]
ALTERNATIVE: [/skillname] — if [condition]
If no existing skill fits, define an ad-hoc approach:
AD-HOC APPROACH:
1. [Step 1]
2. [Step 2]
3. [Step 3]
Step 3: Apply the Thinking
Execute the selected approach against the input.
If a skill was selected:
-> INVOKE: /[skillname] $ARGUMENTS
If using an ad-hoc approach, execute the steps defined above and show the work.
Step 4: Evaluate Output Quality
After the thinking is complete, assess the result.
OUTPUT QUALITY CHECK:
- Does it answer what was actually asked? [yes/no]
- Is it at the right level of depth? [yes/no — too shallow / too deep / right]
- Are there obvious gaps? [list any]
- Would a different thinking type have worked better? [yes/no — which one]
Step 5: Iterate If Needed
If the quality check reveals problems:
ITERATION NEEDED: [yes/no]
REASON: [what's wrong]
SWITCHING TO: [different approach or deeper pass]
If no iteration needed, present the output as the final answer.
Integration
Use with:
/meta-> Get oriented before thinking/sp-> Improve the question before thinking about it/vldt-> Validate the output of your thinking/lrnk-> If the thinking reveals a knowledge gap