# Unknowns Tweakable Plan

> Write implementation plans sorted by the decisions the human is most likely to tweak, with mechanical work buried below. Use when the user asks for a tweakable plan, decision-first plan, plan review artifact, alternatives before sequencing, or wants judgment calls surfaced before implementation.

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

---


# Unknowns Tweakable Plan

## Workflow

Optimize for review, not chronological execution.

Return:

- The top decisions likely to change the plan.
- For each decision: recommended choice, alternative, "choose this if", consequences, and evidence.
- Type/API/data/user-facing impacts.
- Execution sequence after the decisions.
- Mechanical work grouped last.
- Copyable tweaks the user can send back.
- Stop condition: high-impact judgment calls are accepted or edited.

## Rules

Do not lead with routine step order. If a step is obvious plumbing, move it below the decisions. Surface irreversible or expensive-to-change choices early.

Use HTML when toggles or side-by-side alternatives make plan review easier.