| name | clarity-toolkit |
| description | Use when someone is overwhelmed by information, stuck in vague thinking, or spinning in analysis without moving forward. Three modes: cut signal from noise, eliminate vague language, and break overthinking loops. Triggers on: "overwhelmed", "too |
| scenarios | ["I have too much data and can't figure out what matters","My goal is to 'be more strategic' but I don't know what that means","I've been thinking about this for two weeks and I'm going in circles","뭐가 중요한지 모르겠어, 정보가 너무 많아","목표가 너무 모호해서 뭘 해야 할지 모르겠어","계속 생각만 하고 결정을 못하고 있어"] |
| compatibility | {"recommended":["think-tool"],"optional":["sequential-thinking"],"remote_mcp_note":"think-tool이 있으면 세 가지 모드(신호/노이즈·모호성·과잉사고) 중 무엇이 실제 문제인지 정확하게 진단할 수 있습니다. Claude 설정 → MCP Servers에서 remote SSE 엔드포인트를 추가하세요."} |
Clarity Toolkit
When to Use / When Not to Use
Use when:
- Too much information and no clear priority
- A goal or statement can't be acted on because it's too vague
- Analysis is cycling without converging
Not for:
- Choosing between well-defined options (use tradeoff-articulator)
- Identifying cognitive biases (use bias-auditor)
Process
Diagnose the mode first — don't apply all three to every problem.
| Mode | Problem | Symptom |
|---|
| Signal/Noise | Too much input | Conflicting data, too many priorities |
| Vagueness Eliminator | Too little precision | Abstract goals, fuzzy feedback |
| Overthinking Detector | Too much processing | Cycling analysis, seeking "one more" data point |
Mode 1 — Signal/Noise Separator
Run every piece of information through:
- Actionability: Does knowing this change a decision in the next 30 days?
- Source quality: Primary data or someone's interpretation?
- Recency: Still accurate?
- Relevance: Directly bears on the goal?
Signal Stack:
MUST KNOW (changes the decision)
GOOD TO KNOW (context only)
INTERESTING BUT NOISE (log and ignore)
Forcing function for competing priorities: "If I could only act on one thing this week, what has the highest leverage?"
Mode 2 — Vagueness Eliminator
Test: can two people read this and independently know whether they've succeeded? If not, it's vague.
Fix pattern: add metric + current state + target state + timeframe + scope.
- Vague: "Be more strategic."
- Fixed: Ask "What would I see you doing differently? Give me a concrete example from last week."
Operationalization test:
- What would I observe if this were true?
- What would I observe if false?
- What would I measure?
- Who is responsible, by when?
Mode 3 — Overthinking Detector
MCP note: If think-tool is available, use it before diagnosing Mode 3. Confirm this is overthinking, not a genuinely complex problem.
Overthinking signatures:
- Same options considered longer than the decision warrants
- Each new information piece generates a new question
- Gut answer exists but case is being built against it
- Cost of being wrong is low but treated as high
Circuit breakers:
- 10/10/10 test: How do you feel in 10 min / 10 months / 10 years? Same answer = stop analyzing.
- Reversibility test: Is this reversible? If yes — decide now, optimize later.
- Minimum viable answer: What's the smallest decision that enables one concrete action right now?
Output Template
State which mode(s) apply and why. Then:
- Mode 1: Signal stack with explicit noise labeling
- Mode 2: Vague statement rewritten as operational statement (show the transformation)
- Mode 3: Name the overthinking pattern + one circuit breaker + concrete action
Always end with one concrete next step the user can take in the next hour.
What Claude Does / What You Do
| Claude | You |
|---|
| Diagnoses which clarity mode applies | Provide the information, goal, or decision |
| Runs the appropriate filtering/precision/circuit-breaker | Confirm whether the diagnosis fits |
| Rewrites vague goals into operational ones | Commit to the concrete next step |
| Names the overthinking pattern explicitly | Execute, rather than continue analyzing |
Related Skills
tradeoff-articulator — when options are defined but costs need mappingassumption-extractor — when the vagueness is rooted in unexamined premises
