Claude Planning Skills — Design skill for Claude Code
Four interlocking Claude skills: brainstorm a design, plan it, build it in reviewed batches, with a skeptic that questions scope at every gate.
How to install Claude Planning Skills
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open maccydee/claude-planning-skills and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Claude Planning Skills does
Four interlocking Claude skills: brainstorm a design, plan it, build it in reviewed batches, with a skeptic that questions scope at every gate.
Alternatives in Design
- Brainstorm Experiments New — Design lean pretotypes for new product validation 7.8k ★
- Contributing To MiniCode — Thanks for contributing to MiniCode 388 ★
- Brainstorming.SKILL — Collaborative design refinement through Socratic dialogue 262 ★
README
Planning skills for Claude
[](https://docs.claude.com/en/docs/claude-code/skills) [](LICENSE)
Four skills that take a piece of work from a fuzzy idea to shipped code, with a reviewer that keeps asking whether you should be building it at all.

They're built for [Claude Code](https://docs.anthropic.com/en/docs/claude-code) and the skill system: Claude loads a skill when the task matches its description, then follows the instructions inside. These four are designed to hand off to each other, but each works on its own.
The workflow
**`brainstorming`** settles the design before any code exists. It works one question at a time and always leads with a recommendation, so you're validating or redirecting a proposal rather than staring at a blank page. Out comes a `DESIGN.md` that captures the decisions and the trade-offs behind them.
**`writing-plans`** turns that design into an implementation plan a zero-context executor can follow: exact file paths, complete code, test-first cycles, and how to verify each step. The point is that whoever builds it (a fresh session, a teammate, you in a month) shouldn't have to guess. Out comes a `PLAN.md`.
**`executing-plans`** builds the plan in small batches with a review checkpoint after each one, so a wrong turn gets caught while it's cheap to fix instead of twenty tasks later. Every task is test-first and its output is checked for real, not just green.
**`skeptic`** is the reviewer. Its first question is never "is this built well?" but "should this exist at all?". Blunt YAGNI and scope critique, run at any stage: over a design, over a plan's scope, or over what got built before it merges. It ends on a verdict: PROCEED, NEEDS WORK, or RETHINK.
Each stage hands its artifact to the next. Nothing gets built before the skeptic has had a pass.
In
Related Skills
Brainstorm First
Before starting any non-trivial task (>0.5 day, new feature, new UI page, new integration, refactor spanning 5
Design Reviewer
You are a witness gate reviewer with the design lens. Your reviewed content is a
Foundry
Turn a goal into finished, tested and reviewed work. Foundry runs locally and orchestrates Claude Code or Open
Revise
Change an existing GraphCompose template as a new revision — content, assets, theme, layout or a pure refactor
Gate P3
/gate-p3 – Quality Gate 3: Architecture & Design → Planning
Claude Plan With Me Skill
A Claude skill that turns plan mode into a structured, repo-grounded design conversation with senior-engineer-
Related Agents
Metis
Pre-planning gap analyzer. Catches hidden intentions, ambiguities, and AI failure points BEFORE plan generatio
Scope Creep Detector
Scope validation specialist comparing plan against original brainstorm and research to catch feature creep
Fable Research
Use PROACTIVELY for bioscience-research thinking and writing that does NOT design molecules or sequences. MUST