/deep-dive — Deep Investigation Workflow

skills/j-deep-dive/ is the canonical, hand-edited directory for this skill, per CLAUDE.md's "The Canonical Naming Contract" (the E50 reopening of 2026-09-09, which promoted skills/j-deep-dive/ from generated twin to sole canonical form). The j- prefix is there for collision safety — a real directory under a distinct name, so a host tool shipping its own same-named built-in command cannot shadow it (Claude Code's native skill resolution is a literal-string, directory-name-based match; see project/documentation/skill-authoring.md's "Invocation Convention").

⚠️ scripts/generate-j-alias.sh was retired by E50_S14 and no longer exists — there is nothing to run. This file was previously generated from a bare skills/deep-dive/SKILL.md source; E50_S15 deleted that directory. This file is now the sole canonical, hand-edited source for this skill — edit it directly.

Overview

A multi-phase workflow that takes a user's request through structured investigation, collaborative brainstorming, critical scrutiny, and optional solution assessment — producing a refined, saved output document.

Modes

Default mode: challenge

Instructions

Step 1 — Receive the Request

Accept the user's mission, issue, or idea as the request. Note it clearly before proceeding.

Step 2 — Select Mode

If the user has not specified a mode, ask:

"How deep should we go?"

1. Explore — Brainstorm only
2. Challenge — Brainstorm + critical scrutiny (default)
3. Resolve — Full pipeline with solution assessment

If the user is in a hurry or it's obvious from context, default to challenge without asking.

Step 3 — Information Gathering (Judgment Call)

Before brainstorming, assess whether sufficient context exists to proceed productively.

• Check project/documentation/examples/ for related prior work or descriptions
• Consider invoking /examplify if a concept in the request needs clarification
• Ask the user directly if a critical piece of context is missing

Default: proceed. Do not over-investigate. The brainstorm step will surface gaps naturally. Move on unless something fundamental is clearly missing.

Step 4 — Interactive Brainstorm

Invoke /brainstorm with the request and any context gathered in Step 3.

This is an interactive session — the user participates actively. Ask focused questions, challenge assumptions, and explore the design space together.

When the user signals they are satisfied with the brainstorm output, capture the full brainstorm result as the working document and proceed.

Step 5 — Scrutiny Review (challenge and resolve modes)

Summon the scrutiny-agent (agents/scrutiny-agent.md) with the brainstorm output as input.

The scrutiny agent will:

• Produce a structured critical assessment (assumptions, key questions, risks, strengths)
• Save it to ./scrutiny-<slug>.md
• Return the assessment to the scrum-master

Escalation threshold is low. After reading the scrutiny output, escalate to Step 6 if any of the following are true:

• Any risk is rated Med or High severity
• Any key question remains unanswered and is not trivial
• Any core assumption is directly challenged without a clear resolution

If none of the above apply and the mode is challenge, skip Step 6 and proceed to Step 7.

Step 6 — Solution Assessment (resolve mode, or escalated from challenge)

Summon the solution-assessor (agents/solution-assessor.md) in Mode B (chained) — pass the full scrutiny output as input.

The solution assessor will:

• Derive all problems from the scrutiny output
• Assess candidate solutions with effort, risk, and viability
• Save the assessment to ./solution-assessment-<slug>.md
• Return the full assessment to the scrum-master

Step 7 — Amend the Brainstorm Output

The scrum-master amends the captured brainstorm document — does not replace it.

Append a ## 🔍 Deep Dive Synthesis section at the end of the brainstorm document containing:

## 🔍 Deep Dive Synthesis

### Scrutiny Findings
[Brief summary of critical concerns, key questions, and risks identified]
→ Full assessment: `./scrutiny-<slug>.md`

### Solution Paths *(if applicable)*
[Brief summary of recommended solution paths and estimated effort]
→ Full assessment: `./solution-assessment-<slug>.md`

### Open Decisions
[Any unresolved questions or choices the user must make before proceeding]

Keep each section concise — these are signposts to the full documents, not rewrites of them.

Step 8 — Save Output

Determine the correct output directory based on content:

Save the amended brainstorm document as <slug>.md in the chosen directory. Also move any scrutiny/solution-assessment files from the working directory into the same directory for co-location.

Inform the user of all saved file paths.

Step 9 — Offer Next Steps

Ask the user:

"Would you like to commit this to the board?"

1. Yes — proceed with /todo to create board items from the output
2. No — close the deep-dive session