Turn "build this" into a real plan
The Super Planner Skill for Claude Code
Most planning prompts are a giant wall of text that treats a one-line copy fix the same as a new payment integration. This drop-in Claude Code skill doesn't. It classifies the task into Quick / Standard / Deep first, then runs only the ceremony that tier needs — so a small change gets a one-screen plan and a real build gets the full research → refine → goal-contract → plan → execute → verify loop. It's the planner I actually run on my own projects. Free, MIT, and on GitHub.
Pay what you want (including $0) · one skill folder · no dependencies · MIT
The problem
One giant planning prompt fits nothing well
The same wall of text either drowns a tiny fix in ceremony or under-plans a real build. And most prompts happily invent an API or a command they never verified.
▌ A one-size planning paste
- Treats a copy tweak and a payment integration the same way
- Ten gates of ceremony on a task that needed one line
- Invents commands, paths, or flags it never checked
- Calls it "done" with no check behind the claim
▌ The Super Planner Skill
- Sizes the task first — Quick, Standard, or Deep — and runs only that
- Routes research: official docs for facts, a search for market signal
- Locks a goal with numbered, individually checkable success criteria
- Is told not to report "done" without a passing check for each one
What it does
How it works
It states the tier it picked, runs that tier, pauses at your approval, then guides the build and verification. Opt-in only — it never fires on a casual question.
One-screen plan
A small, known change. Straight to a mini-plan: one-line goal, files touched, the exact change, the verify command. One approval, build, verify. No research, no gates.
Light path, one gate
A real feature in a known codebase. Docs-verify research + a single combined approval gate, then build and verify — without the full deep ceremony.
The full pipeline
A new project, an unfamiliar API, or anything touching auth / payments / third-party data. Full research → refine → goal-contract → routing → plan → execute → verify.
Checkable, not vibes
Numbered success criteria, each individually checkable, plus a runnable definition of done — and every build slice traces back to one of those criteria.
Verify or say so
No API, command, path, price, or flag from memory. It verifies against your repo or current docs, or says it can't. A stated gap beats a confident guess.
What's in the box
What's inside
- ✓The skill — a single Markdown file you drop into ~/.claude/skills/
- ✓START-HERE.html — a friendly install + use walkthrough, copy button on the command
- ✓README (Markdown + plain text) and an MIT license
- ✓The same thing on GitHub — star it, fork it, open an issue
- ✓No dependencies, nothing to build — Claude Code picks it up next session
⚠ Requirements — please read before buying
- Claude Code — the skill uses its /command and skill loader
- Nothing else. The skill is a single Markdown file, no packages, no runtime
- Some phases mention optional tools (docs-fetch, web-search, deep-research); if one isn't there, it uses a direct source when it can, or says it can't verify that part
- It's opt-in only — plain "help me plan X" won't trigger it; you invoke /jd-super-plan
Free, MIT, 30-second install
Stop pasting a wall of text. Plan on purpose.
One command that sizes the task, routes the research, locks a checkable goal, and won't wave "done" through. Free — pay what you want, including nothing.
Instant download · 30-day Gumroad refund policy applies
Setting up Claude Code from scratch? Start with the Claude Code Setup Playbook — $29.
See all my Claude Code tools → expressive446.gumroad.com