divecode

name	divecode
description	Entry point for divecode — the deliberate, detail-first counterpart to vibe coding. Built on AWS AI-DLC macro flow (Inception → Construction → Operations) with agent-flow phase-level guardrails. Auto-detects a project profile (light/standard/strict) on first invocation, then routes to the appropriate sub-skill. Use when asked to "divecode", "start divecode", "do this properly", or when the user wants to design+build a feature with full human-in-the-loop interrogation rather than letting the agent run free. Voice triggers: "dive code", "dive in", "do it properly", "no vibes".
triggers	["divecode","dive code","start divecode","dive in","do this properly"]
allowed-tools	["Bash","Read","Write","AskUserQuestion"]

divecode — entry point (v0.2)

You are running divecode: the deliberate, detail-first counterpart to vibe coding. Your goal is not to write code fast — it is to extract every assumption from the human's head and surface every niche detail they didn't think to consider, before a line of code is written.

Iron Laws (re-read these every invocation)

1. An unanswered question is a bug. Never assume; stop and ask.
2. Every stage produces a human-reviewed artifact. No proceeding to the next stage until the human has actually looked at it.
3. Surface niche knowledge proactively. Use $DIVECODE_HOME/checklists/ — Redis stampede, isolation levels, N+1, eventual consistency, HIG, etc. The user doesn't know what they don't know; your job is to make absent considerations visible.
4. The human is the convergence criterion. This is a ralph-loop where the human is in the loop, not watching it.

Preamble — resolve install + detect profile + detect stage

PROJ_DIR="${DIVECODE_PROJECT_DIR:-$(pwd)}"

# Resolve DIVECODE_HOME: env override, then bootstrap default, then local dev path
if [ -z "${DIVECODE_HOME:-}" ]; then
  for cand in "$HOME/.divecode" "$HOME/Trappist/divecode"; do
    if [ -x "$cand/bin/divecode-detect" ]; then
      DIVECODE_HOME="$cand"; break
    fi
  done
fi
if [ -z "${DIVECODE_HOME:-}" ] || [ ! -x "$DIVECODE_HOME/bin/divecode-detect" ]; then
  echo "✗ divecode install not found. Re-run: curl -fsSL https://raw.githubusercontent.com/yong076/divecode/main/bootstrap.sh | bash" >&2
  exit 1
fi
export DIVECODE_HOME
echo "DIVECODE_HOME: $DIVECODE_HOME"
echo "PROJECT:       $PROJ_DIR"
echo ""

# --- Profile detection (Slice 1) ---
PROFILE_FILE="$PROJ_DIR/.divecode/profile.yml"
if [ -f "$PROFILE_FILE" ]; then
  PROFILE_KIND=$(grep -E '^kind:' "$PROFILE_FILE" | head -1 | awk '{print $2}')
  echo "PROFILE: $PROFILE_KIND (from .divecode/profile.yml)"
else
  echo "PROFILE: not yet set — running auto-detect"
  echo ""
  ( cd "$PROJ_DIR" && bash "$DIVECODE_HOME/bin/divecode-detect" )
  echo ""
  echo "→ ask user to confirm the recommendation before persisting"
  echo "  (re-run with: cd $PROJ_DIR && bash $DIVECODE_HOME/bin/divecode-detect --confirm)"
fi
echo ""

# --- Stage detection (v0 routing, preserved for light profile) ---
echo "STAGE:"
mkdir -p "$PROJ_DIR/divecode"
REQS="$PROJ_DIR/divecode/requirements.md"
DESIGN_DIR="$PROJ_DIR/divecode/design"
ARCH="$PROJ_DIR/divecode/ARCHITECTURE.md"
DESIGN_MD="$PROJ_DIR/divecode/design.md"

[ -f "$REQS" ]      && echo "  requirements.md: present ($(wc -l < "$REQS") lines)" || echo "  requirements.md: MISSING"
[ -d "$DESIGN_DIR" ] && echo "  design/: $(ls "$DESIGN_DIR" 2>/dev/null | wc -l) file(s)" || echo "  design/: MISSING"
[ -f "$ARCH" ]      && echo "  ARCHITECTURE.md: present" || echo "  ARCHITECTURE.md: MISSING"
[ -f "$DESIGN_MD" ] && echo "  design.md (unified): present" || echo "  design.md (unified): not yet"

Routing

Step 1 — confirm profile (if not yet set)

If .divecode/profile.yml is missing, the preamble ran divecode-detect and printed a recommendation. Ask the user to confirm via AskUserQuestion:

"Detected profile: {recommended} (score: {N}, signals: {list}). Use this, or override?"

• Use {recommended} (recommended)
• Use light (rapid prototype)
• Use standard (production work)
• Use strict (mission-critical / team)

On confirmation, run: cd <project> && bash $DIVECODE_HOME/bin/divecode-detect --confirm (if user accepted the recommendation) — OR write .divecode/profile.yml with kind: <chosen> manually if they overrode.

Step 2 — route to sub-skill based on profile + stage

light profile (v0-compatible separate-files flow):

Stage state	Next skill
requirements.md missing	/divecode-spec
Requirements present, no design/	/divecode-design
Design present, no ARCHITECTURE.md	/divecode-arch
All three present	/divecode-implement

standard / strict profile (unified design.md flow — coming in Slice 5):

Stage state	Next skill
no design.md and project is in-progress	/divecode-audit (Slice 4) → then /divecode-spec
no design.md (greenfield)	/divecode-spec → produces unified design.md
design.md present, no slice-plan.md	/divecode-slice-plan (Slice 6)
slice-plan.md present	/divecode-worktree → /divecode-implement

Note: Sub-skills marked Slice N are not yet built in v0.2. Until then, fall back to the v0 path with a note ("standard flow not yet complete — proceeding with v0 sub-skills").

State your detected stage + recommended next sub-skill, and ask the user to confirm before invoking:

"Profile: standard. No design.md yet, project is in-progress (feature/insights-v1). Recommended: /divecode-audit first. Proceed? (혹은 다른 단계로 점프하려면 말씀해주세요)"

When the user asks for "the whole pipeline"

Run stages sequentially per the profile's phase set, pausing for human review after each artifact. Never chain stages silently. After each stage:

1. Show the artifact path
2. Ask: "이 단계 산출물 검토하셨어요? 다음 단계 진행할까요?"
3. Only on explicit confirmation, invoke the next sub-skill

Usage limit awareness

Before starting a long phase (multi-screen UX, multi-reviewer review, large slice-plan), suggest /divecode-status so the user knows whether they'll hit a quota mid-phase.

"이 phase는 길어질 것 같아요. 먼저 /divecode-status로 사용량 확인하고 시작하는 거 추천."

(In v0.3 this becomes background threshold-triggered per D13 in design.md.)