# Decision Framework

> Decision Framework

- Skill: `sethdford/decision-framework` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add sethdford/decision-framework`
- Raw SKILL.md: https://api.skillmd.com/api/skills/sethdford/decision-framework/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: sethdford (https://skillmd.com/u/sethdford)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/sethdford/decision-framework

---

# Decision Framework

Choose among options using explicit criteria, foresight, and reversibility—not gut feel alone. Combine scoring with “and then what?” and one-way vs two-way door tests before committing.

## When to Use
- Multiple viable paths, high stakes, or disagreement about what “best” means.
- You need a documented rationale others can audit or revisit later.

## Workflow
1. **Frame** the decision: decision owner, deadline, success definition, and non-goals.
2. **List options** (including “do nothing” / status quo when realistic).
3. **Criteria & weights**: name 3–7 criteria; assign weights that sum to 100%; justify weights briefly.
4. **Weighted scoring matrix**: score each option per criterion (use a consistent scale); compute totals; sanity-check winners against intuition.
5. **Second-order thinking**: for top options, ask “and then what?” twice—who reacts, what breaks, what new constraints appear?
6. **Reversibility test**: label each option one-way door (hard to undo) vs two-way door (cheap to revert); favor experimentation for two-way doors.
7. **Regret minimization**: which choice minimizes long-run regret if your medium-term assumptions are wrong?
8. **Opportunity cost**: what does the leading option foreclose? Name the next-best use of the same time, money, or focus.
9. **OODA skim** (when time is short): Observe facts, Orient to constraints, Decide on a default, Act with a review trigger.
10. **Decide & document**: chosen option, key scores, second-order notes, reversibility, and what would trigger a rethink.

## Examples
**Example 1:** Picking a vendor: weighted matrix on reliability, support, cost, lock-in; second-order effects on hiring and integrations; one-way door if data migration is painful.

**Example 2:** Architecture fork: score maintainability vs speed to ship; pre-mortem on operational load; treat behind-feature-flag rollout as a two-way door with a rollback plan.