# Exit Or Double Down

> This skill should be used when the user asks "should we exit this business," "is this worth continuing to fund," "do a portfolio pruning review," or needs an explicit exit/hold/double-down call on a unit — with sunk-cost reasoning named directly when it appears.

- Skill: `daddia/exit-or-double-down` (Agent Skill)
- Install (CLI): `npx skillmds@latest add daddia/exit-or-double-down`
- Raw SKILL.md: https://api.skillmd.com/api/skills/daddia/exit-or-double-down/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: daddia (https://skillmd.com/u/daddia)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/daddia/exit-or-double-down

---

# Exit or Double Down

## When to use

Portfolio pruning — explicit exit/hold/double-down with blank-page test first and sunk-cost reasoning named when it appears.

## What this skill does not do

- **Does not treat sunk cost as forward evidence** — quoted and set aside.
- **Does not default to HOLD** — requires justification + kill criteria.
- **Does not invent financials** — INPUT NEEDED.
- **Does not name vague approver** — routing per profile.

## Preconditions

| Input | If missing |
|---|---|
| Org + corporate-strategy profiles | practice-setup or `[PROVISIONAL]` |
| Unit scope and decision forum | Ask |

## Provisional mode

Generic exit criteria; `[PROVISIONAL]` on thresholds.

## Trust spine

- **Strategic advice vs. support (mandatory):** CALL is recommendation; named forum approves.
- Per trust-conventions; GATE before board/IC finals.

## Workflow

1. Orient — unit, forum, evidence, time pressure.
2. **Blank-page test first** — enter today at this investment?
3. Sunk-cost detection — quote patterns, set aside.
4. Exit criteria check — structural decline, capital destruction, growth odds.
5. Double-down criteria — position, leadership path, under-resourcing.
6. CALL: EXIT | HOLD | DOUBLE DOWN — forward rationale only.
7. Kill criteria if HOLD.
8. Approval routing per profile.
9. GATE if board-ready.

## Output format

```
CONFIDENCE: [...]
UNIT: [...]
BLANK-PAGE TEST: [...]
SUNK-COST CHECK: [...]
EXIT CRITERIA: [table]
DOUBLE-DOWN CRITERIA: [table]
CALL: [...]
KILL CRITERIA (if HOLD): [...]
APPROVAL ROUTING: [...]
```

## Worked example

**Input:** Legacy print unit, "we've spent $200M already" cited.

**Excerpt:** Blank-page: no. Sunk-cost quoted and set aside. EXIT criteria met on structural decline; CALL EXIT `[review]`.

## Quality checks before delivering

- [ ] Blank-page before other evidence
- [ ] Sunk cost named if present
- [ ] Hold justified with kill criteria
- [ ] Approver named

## Propose profile update

When a stable convention surfaces during this run (thresholds, naming, tone, output format, or recurring corrections), **propose a profile update**: show the exact diff against `~/.claude/plugins/config/claude-for-strategy/corporate-strategy/CLAUDE.md` (org-wide facts go to `org-profile.md`), ask for confirmation, and write only on yes. Only `/corporate-strategy:practice-setup` auto-applies a full profile write.

## Outputs

Follows plugin `CLAUDE.md` § Outputs. Next: IC exit case, `allocate-resources` for double-down sizing.