STOP — DO NOT READ THIS FILE. You are already reading it. This prompt was injected into your context by Claude Code's plugin system. Using the Read tool on this SKILL.md file wastes ~7,600 tokens. Begin executing Step 1 immediately.

/pbr:new-project — Project Initialization

References: @references/questioning.md, @references/ui-brand.md

You are the orchestrator for /pbr:new-project. This skill initializes a new Plan-Build-Run project through deep questioning, optional research, requirements scoping, and roadmap generation. Your job is to stay lean — delegate heavy work to Task() subagents and keep the user's main context window clean.

Context Budget

Reference: skills/shared/context-budget.md for the universal orchestrator rules. Reference: skills/shared/agent-type-resolution.md for agent type fallback when spawning Task() subagents.

Additionally for this skill:

• Minimize reading subagent output — read only summaries, not full research docs
• Delegate all analysis work to subagents — the orchestrator routes, it doesn't analyze

Step 0 — Immediate Output

Before ANY tool calls, display this banner:

╔══════════════════════════════════════════════════════════════╗
║  PLAN-BUILD-RUN ► STARTING PROJECT                           ║
╚══════════════════════════════════════════════════════════════╝

Then proceed to Step 1.

Multi-Session Sync

Before any phase-modifying operations, this skill acquires a claim on the project:

acquireClaim(planningDir, sessionId)

Where planningDir is the .planning/ directory and sessionId is the current session identifier. If the claim fails (another session owns it), display: "Another session owns this project. Use /pbr:progress to see active claims."

On completion or error, release the claim:

releaseClaim(planningDir, sessionId)

Prerequisites

• Working directory should be the project root
• No existing .planning/ directory (or user confirms overwrite)

Orchestration Flow

Execute these steps in order. Each step specifies whether it runs inline (in your context) or is delegated to a subagent.

Step 1: Detect Brownfield (inline)

CRITICAL — Run init command FIRST before any Glob calls or manual file checks:

node plugins/pbr/scripts/pbr-tools.js init begin

Store the JSON result as blob. This single call replaces multiple Glob/filesystem checks with a pre-computed payload:

• blob.has_planning — whether .planning/ directory exists
• blob.has_existing_code — whether brownfield indicators were found
• blob.brownfield_indicators — array of detected indicators (package.json, src/, etc.)
• blob.has_git — whether .git directory exists
• blob.existing_phases — count of existing phase directories
• blob.state — existing STATE.md frontmatter (or null if no project)
• blob.config — existing config.json contents (or null)

Cross-platform note: The init command handles all filesystem checks cross-platform. No Glob or Bash file discovery needed.

If blob.has_existing_code is true: CRITICAL -- DO NOT SKIP: Present the following choice to the user via AskUserQuestion before proceeding: Use the yes-no pattern from skills/shared/gate-prompts.md: question: "This looks like an existing codebase. Run /pbr:map-codebase to analyze what's here first?" options: - label: "Yes, scan" description: "Run /pbr:map-codebase first to analyze existing code" - label: "No, begin" description: "Proceed with /pbr:new-project on top of existing code"

• If user selects "Yes, scan": suggest /pbr:map-codebase and stop
• If user selects "No, begin": proceed to Step 2

If blob.has_planning is true: CRITICAL -- DO NOT SKIP: Present the following choice to the user via AskUserQuestion before proceeding: Use the yes-no pattern from skills/shared/gate-prompts.md: question: "A .planning/ directory already exists. This will overwrite it. Continue?" options: - label: "Yes" description: "Overwrite existing planning directory" - label: "No" description: "Cancel — keep existing planning"

• If user selects "No": STOP IMMEDIATELY. Do not ask again. Do not proceed to Step 2. End the skill with this message:

  Keeping existing .planning/ directory. Use `/pbr:progress` to see current project state, or `/pbr:plan-phase` to continue planning.
  

  Do NOT re-prompt the same question or any other question. The skill is finished.
• If user selects "Yes": proceed (existing directory will be overwritten during state initialization)

Step 2: Deep Questioning (inline)

Reference: Read references/questioning.md for technique details.

Have a natural conversation to understand the user's vision. Do NOT present a form or checklist. Instead, have a flowing conversation that covers these areas organically:

Required context to gather:

1. What they want to build — The core product/feature/system
2. Problem being solved — Why does this need to exist? Who is it for?
3. Success criteria — How will they know it works? What does "done" look like?
4. Existing constraints — Technology choices already made, hosting, budget, timeline, team size
5. Key decisions already made — Framework, language, architecture preferences
6. Edge cases and concerns — What worries them? What's the hardest part?

Conversation approach:

• Start broad: "What are you building?"
• Go deeper on each answer: "What does that mean exactly?" "Show me an example."
• Surface assumptions: "Why do you assume that?" "Have you considered X?"
• Find edges: "What happens when...?" "What about...?"
• Reveal motivation: "Why does that matter?"
• Avoid leading questions — let the user define their vision

Keep going until you have:

• A clear, concrete understanding of what they want to build
• At least 3 specific success criteria
• Known constraints and decisions
• A sense of complexity and scope

Anti-patterns:

• DO NOT present a bulleted checklist and ask them to fill it in
• DO NOT ask all questions at once — have a conversation
• DO NOT assume technologies — let them tell you
• DO NOT rush — this is the foundation for everything that follows

Step 2.5: Fast-Path Offer (inline)

Before asking any workflow preference questions, offer the user a quick-start option:

Use AskUserQuestion: question: "How do you want to configure this project?" header: "Setup mode" options: - label: "Quick start" description: "Use all defaults — model balanced, depth standard, interactive mode, parallel on. Writes config in seconds." - label: "Custom setup" description: "Walk through model selection, features, and preferences step by step."

If user selects "Quick start":

• Tell the user: "Tip: Agents set to inherit use your Claude Code session model. For cost savings, run on Sonnet -- orchestration and inherit agents work well at that tier. See references/model-profiles.md for details."
• Write .planning/config.json immediately using the default config below (no further questions):

  {
    "version": 2,
    "context_strategy": "aggressive",
    "mode": "interactive",
    "depth": "standard",
    "context_window_tokens": 200000,
    "features": {
      "structured_planning": true,
      "goal_verification": true,
      "integration_verification": true,
      "context_isolation": true,
      "atomic_commits": true,
      "session_persistence": true,
      "research_phase": true,
      "plan_checking": true,
      "tdd_mode": false,
      "status_line": true,
      "auto_continue": false,
      "auto_advance": false,
      "team_discussions": false,
      "inline_verify": false
    },
    "models": {
      "researcher": "sonnet",
      "planner": "inherit",
      "executor": "inherit",
      "verifier": "sonnet",
      "integration_checker": "sonnet",
      "debugger": "inherit",
      "mapper": "sonnet",
      "synthesizer": "haiku"