Skip to content

/plan

Comprehensive interview for planning new features or changes. Use when the user wants thorough upfront planning before implementation, or when starting work on a complex feature.

shell
$ npx -y skills add bendrucker/claude --skill plan --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.
  • You can call itInvoke it directly when you want it.
  • Slash command/plan
How auto-invocation works

Context preview

The summary Claude sees to decide when to auto-load this skill.

Comprehensive interview for planning new features or changes. Use when the user wants thorough upfront planning before implementation, or when starting work on a complex feature.

SKILL.md

plan.SKILL.md
name: interview:plan
description: |
  Comprehensive interview for planning new features or changes. Use when the user wants thorough upfront planning before implementation, or when starting work on a complex feature.

Planning Interview

Interview the user about their requested change via the `AskUserQuestion` tool.

Rules

  • Every question goes through `AskUserQuestion`. Never list questions in chat prose.
  • Up to 4 questions per call, batched by topic. The cap is per call, not per interview: chain calls until the plan is solid.
  • Refine the next batch based on prior answers. Do not pre-plan all questions.

Open-Ended Questions

Even when the answer space feels open, sketch 2-4 plausible options. The user gets "Other" automatically (do not add it). If you cannot name two distinct options, read more code or ask a more foundational question first.

When you have a recommendation, put it first and append "(Recommended)". For visual or code comparisons, use the `preview` field on options.

Topics

Adapt to context. Skip obvious answers. Challenge implicit decisions.

  • Requirements: what it should and shouldn't do
  • Technical implementation: architecture, data flow, dependencies
  • UI/UX: interactions, feedback, error states
  • Constraints: performance, compatibility, security, accessibility
  • Tradeoffs and scope boundaries
  • Validation: how we know it works

Output

The interview informs the conversation. Do not write to files unless the user requests it.

Read more
Read it on GitHub ↗
Ships withbendrucker-claude

My personal plugin marketplace for Claude Code, Anthropic's AI coding assistant.

Get the whole plugin, auto-invoked
Stats
15
Stars
0
Views
1
Forks
Active
Maintenance
TypeScript
Language
MIT
License
14h ago
Last commit
1y ago
Created

Repo: bendrucker/claude