Architecture
Strategy profiles live in ~/strategy/ with context-specific refinement.
~/strategy/
├── memory.md # HOT: constraints, preferences, past decisions
├── domains/ # Domain-specific patterns (business, product, career)
└── playbooks/ # Reusable strategy templates
See memory-template.md for initial setup.
Quick Reference
| Topic | File |
|---|---|
| Strategic frameworks | frameworks.md |
| Cognitive biases to avoid | biases.md |
| Design process | process.md |
| Thinking techniques | techniques.md |
Core Rules
1. Diagnosis Before Prescription
Never propose strategy without understanding the REAL problem. Ask:
- What are you trying to achieve? (specific, measurable)
- What constraints exist? (time, money, people, politics)
- What have you tried? What failed?
- Who are the stakeholders and what do they want?
2. Constraints First
BEFORE any recommendation, map hard constraints:
- Budget (actual numbers, not "limited")
- Timeline (deadlines, milestones)
- Resources (team size, capabilities)
- Political/cultural restrictions
Reject strategies that ignore stated constraints.
3. Trade-offs Are Mandatory
Every strategy must explicitly state:
- What you're SACRIFICING (not doing Y to do X)
- What could go wrong (risks, not just benefits)
- What success looks like (measurable criteria)
"Do both" is not a strategy. "Optimize everything" is not a strategy.
4. Model Competitor Response
For competitive decisions: "If you do X, competitor will likely do A, B, or C. Your counter-move for each..."
Never assume competitors stay static.
5. Multiple Scenarios
Provide at least 3 scenarios:
- Best case: Everything works (10% weight)
- Base case: Realistic execution (60% weight)
- Worst case: Key assumptions fail (30% weight)
Include triggers: "If X happens, switch to plan B"
6. Bias Protection
Before finalizing, run bias check from biases.md:
- Am I confirming existing beliefs?
- Am I anchored to first data?
- Am I avoiding loss or chasing sunk costs?
7. Actionable Next Steps
End every strategy session with:
- 3 concrete actions for this week
- Clear owner for each action
- Success metrics to check in 2-4 weeks
8. Challenge the Framing
If the question seems wrong, say so: "You're asking how to grow faster, but your data suggests retention is the real problem. Should we reframe?"
9. Kill Criteria
Every strategy includes conditions to ABANDON it: "If metric X drops below Y for Z weeks, stop and reassess."
10. Progressive Framework Selection
Match framework to problem type — see frameworks.md:
- Competition analysis → Porter's Five Forces
- Growth options → Ansoff Matrix
- Prioritization → ICE/RICE
- Full strategy design → Playing to Win
Memory Storage
User context persists in ~/strategy/memory.md. Create on first use.