PCV: Plan-Construct-Verify

What is PCV?

PCV is a skill for Claude Code that adds structured planning discipline to complex projects. It separates deciding what to build from building it, ensuring you make key decisions explicitly before any code is written, any document is drafted, or any design is rendered.

Without PCV, Claude Code makes dozens of implicit decisions — which packages to use, how components connect, what interfaces look like, how errors are handled. Many are reasonable in isolation but don’t match what you intended. You discover this after the work is done, leading to costly rework or starting over.

PCV surfaces those decisions during a structured planning phase, records your choices in an auditable trail, then executes the approved plan precisely.

When to Use PCV

Prerequisites

Claude Code must be installed and working on your machine. PCV is a Claude Code skill — it runs inside Claude Code, not separately. If you don’t have Claude Code yet, install it first: Claude Code documentation.

Installation

1. Download the bootstrap file.
   Save bootstrap.md to any convenient location on your machine.
2. Give it to Claude Code.
   Start a Claude Code session and say:
   Read [path-to-bootstrap.md] and follow the installation instructions inside it.
3. Confirm the installation.
   Claude Code will install the skill files and agent definitions and report success. You’re ready to go.

Quick Start

1. Create a project directory and cd into it in your terminal.
2. Start Claude Code in that directory.
3. Type /pcv — PCV scaffolds the workspace including idea.md, charge.md (template), and a permissions file.
4. Describe your idea in idea.md — what you’re building, any constraints, what success looks like. Tell Claude you’re ready.
5. Review the generated charge. PCV reads your idea, presents a draft charge in chat, asks clarifying questions, then writes the finalized charge.md on your approval.
6. Planning begins automatically. PCV asks clarification questions (one at a time), runs an adversarial review, and presents plans for your approval.
7. Approve the plans. PCV builds per the approved plan, verifies the result, and offers optional hands-on acceptance testing before closing out.

PCV uses specialized agents for context isolation and token efficiency: a research agent analyzes prior work, a builder agent constructs each component in its own context, a verifier agent checks deliverables, and a critic agent reviews plans adversarially. Agent models are adaptively configured per project — PCV proposes economy or full configurations based on project complexity, and you can adjust at any time. See the Workflow Guide for details on the subagent architecture, adaptive configuration, multi-phase projects, PCV Lite, deployment, project summaries, token efficiency, all four deliverable patterns, and troubleshooting tips.

Four Deliverable Patterns

PCV adapts its planning, construction, and verification to the type of deliverable: