give-plan

Create a reviewable implementation plan from repository files before coding.

113|15|Updated Mar 3, 2026
One-click install
npx skills add https://github.com/avibebuilder/claude-prime --skill give-plan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: give-plan
Source: https://github.com/avibebuilder/claude-prime/tree/main/.claude/skills/give-plan
Command: npx skills add https://github.com/avibebuilder/claude-prime --skill give-plan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Projects and teams often start coding before a clear, reviewable specification exists, which leads to rework, inconsistent implementations, hidden assumptions, and missed validation criteria. This Skill ensures work is paused to produce an inspectable plan that documents scope, tradeoffs, and validation before any code is written.

Core Features & Use Cases

  • Clarify goals with focused questions: Ask only what materially changes the plan so the artifact is precise and reviewable.
  • Ground recommendations in the repository: Read relevant files, separate confirmed facts from inferences, and call out unknowns.
  • Choose the right artifact size: Produce inline prose for tiny changes, timestamped plan files for medium work, or phased directories for large rollouts.
  • Provide validation and rollout guidance: List affected files/systems, dependencies, sequencing, validation checkpoints, rollback steps, risks, and mitigations.
  • Use Cases: drafting a migration rollout, proposing an architecture decision for team review, phasing a multi-file feature, or writing a spec before an assigned implementation.

Quick Start

Create a phased implementation plan that summarizes the problem, recommended approach, affected files, phases, validation steps, risks and mitigations, and explicit approvals needed.

Frequently Asked Questions about give-plan

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a reviewable implementation plan before coding begins?

To create a reviewable implementation plan, you outline recommended changes, read repository files to distinguish confirmed facts from inferences, list affected files, and define validation criteria before writing any code. This ensures scope, tradeoffs, and validation are documented and approved upfront.

What is an architecture decision record and when do I need a written spec?

A written spec or architecture decision document is needed when starting phased features, system migrations, or scoped implementations. It prevents rework and hidden assumptions by pausing development to produce an inspectable plan that documents scope, tradeoffs, and validation checkpoints for team review.

How do I plan a database migration rollout with validation and rollback steps?

Planning a migration rollout involves creating a phased implementation plan that lists affected files and systems, dependencies, sequencing, validation checkpoints, rollback steps, risks, and mitigations. The plan pauses for explicit user approval before implementation begins to ensure safe execution.

Can I use inline prose for small code changes instead of a full plan directory?

Yes, you can use inline prose for tiny changes instead of a full plan directory. The approach scales artifact size based on scope: inline prose for small changes, timestamped plan files for medium work, and phased directories for large multi-file rollouts.

How do I separate confirmed facts from inferences when reviewing repository files?

Separating confirmed facts from inferences involves reading relevant repository files directly and explicitly calling out unknowns in the implementation plan. This grounds recommendations in the actual codebase, ensuring the proposed architecture decisions and validation criteria are based on verified project state.

What is the best way to phase a multi-file feature implementation?

The best way to phase a multi-file feature implementation is to generate a phased directory structure that summarizes the problem, recommended approach, affected files, validation steps, risks, and mitigations. This pauses coding to provide an inspectable artifact for explicit team approval.