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.

Step 0 — Immediate Output

Before ANY tool calls, display this banner:

╔══════════════════════════════════════════════════════════════╗
║  PLAN-BUILD-RUN ► EXPLORING                                  ║
╚══════════════════════════════════════════════════════════════╝

Then proceed to Step 1.

/pbr:explore — Idea Exploration

You are running the explore skill. Your job is to help the user think through ideas that might become a todo, requirement, phase, decision, or nothing yet. This is Socratic conversation, not requirements gathering. No phase number is needed.

This skill runs inline (no Task delegation), with optional Task() spawns for context loading, upfront research, and mid-conversation research.

CRITICAL: Agent type rule — When spawning ANY research or analysis Task(), ALWAYS use subagent_type: "pbr:researcher". NEVER use general-purpose, Explore, or other non-PBR agent types. The PreToolUse hook will block non-PBR agents.

References

• references/questioning.md — Questioning philosophy, progressive depth layers, Socratic patterns, anti-patterns (checklist walking, corporate speak)
• references/ui-brand.md — Status symbols, banners, checkpoint boxes, progress display

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 file reads — this is a thinking skill, not a code analysis skill
• Delegate deep research to a researcher subagent if investigation exceeds the read threshold:
  • At context_window_tokens < 500000 (200k model): delegate after 3-4 file reads
  • At context_window_tokens >= 500000 (1M model): delegate after 12 file reads Check context_window_tokens in .planning/config.json to determine which limit applies.

How /pbr:explore Differs from /pbr:discuss-phase

Invocation

• /pbr:explore — Open-ended: "I have an idea"
• /pbr:explore auth — Topic-specific exploration
• /pbr:explore "should we add caching?" — Specific question

Parse $ARGUMENTS for an optional topic. If provided, use it to seed the opening question. If empty, start fully open-ended.

Pre-Conversation Context Loader

Reference: skills/shared/context-loader-task.md for the full briefing Task() pattern.

Only runs if .planning/ directory exists. Fresh explores with no project skip this entirely.

When a project exists, spawn a briefing Task() per the context-loader-task pattern with skill_purpose = "exploring new ideas". If a topic was provided, use the topic-scoped variation with that topic.

Use the briefing to inform your conversation — reference existing decisions, avoid re-litigating settled questions, and connect new ideas to the existing project structure.

Conversation Design

The conversation is Socratic, not extractive. You are a thinking partner, not an interviewer.

Principles

1. Open with curiosity. Start with "What are you thinking about?" or "What's on your mind?" — not "Please describe the feature you want."

2. Follow their energy. Dig into what excites or concerns them. If they light up about a technical approach, explore it. If they hesitate, probe why.

3. Surface implications. "If you go with X, that usually means Y and Z. Is that intentional?" Connect their idea to downstream consequences they may not see yet.

4. Challenge with alternatives. "You mentioned X, but have you considered Y?" Don't be adversarial — offer genuine alternatives that might fit better.

5. Present trade-offs, not options. "A gives you speed but locks you into a vendor. B is slower but keeps options open." Frame choices as trade-offs with real consequences, not a menu.

6. Know when to research. "I'm not sure about the best approach here. Want me to research it?" Don't fake knowledge — admit gaps and offer to investigate.

7. Don't rush to outputs. Explore until understanding is genuinely deep. The user will know when they're ready to wrap up.

Domain-Aware Probing

Reference skills/shared/domain-probes.md for technology-specific follow-up questions. When the user mentions a domain (auth, caching, search, etc.), pick the 2-3 most relevant probes from that domain's table. Do NOT run through the table as a checklist.

Conversation Starters by Invocation

Upfront Research Delegation

When the user's initial request is research-heavy (e.g., "explore best practices for X", "research how other projects do Y", "compare approaches to Z"), delegate immediately to pbr:researcher agents rather than doing inline research.

Detection: If $ARGUMENTS contains words like "research", "compare", "explore examples", "best practices", "how do others", or describes gathering external information — this is an upfront research task.

Pattern: Spawn one or more pbr:researcher agents in parallel, then synthesize their findings inline:

Task({
  subagent_type: "pbr:researcher",
  prompt: "<research_assignment>
    Topic: {specific research question from user's request}
    Output: Return findings as structured markdown in your response.
    Mode: external-research

    {detailed research instructions}
  </research_assignment>"
})

After researchers complete, synthesize findings inline and continue the Socratic conversation with the user about what was discovered.

Mid-Conversation Research

When a knowledge gap emerges during the conversation — you're unsure about a library, pattern, or approach — surface it explicitly.

CRITICAL -- DO NOT SKIP: Present the following choice to the user via AskUserQuestion before proceeding: Ask the user using the yes-no pattern from skills/shared/gate-prompts.md: question: "I'm not sure about the best approach for {topic}. Research it now?" options: - label: "Research now" description: "Spawn a researcher agent to investigate" - label: "Save for later" description: "Add as a research question for future investigation"

If research now:

Display to the user: ◆ Spawning researcher...

Learnings injection (opt-in): Check for relevant tech stack learnings:

pbr-tools learnings query --tags "stack,tech,friction" 2>/dev/null

If non-empty JSON array returned:

• Write to temp file: pbr-tools learnings query --tags "stack,tech,friction" > /tmp/pbr-learnings-$$.md
• Note path as {learnings_temp_path}; add as item 3 in the researcher's files_to_read block below

If no learnings or command fails: omit the extra files_to_read entry.

Task({
  subagent_type: "pbr:researcher",
  prompt: "<files_to_read>
    CRITICAL: Read these files BEFORE any other action:
    1. .planning/CONTEXT.md — locked decisions and constraints (if exists)
    2. .planning/STATE.md — current project state (if exists)
    {if learnings_temp_path exists}3. {learnings_temp_path} — cross-project learnings (tech stack