jdi-new

name	jdi-new
description	Entry point for new project. Runs research + asker, generates PROJECT.md + ROADMAP.md. --auto (alias --yolo) skips all questions — the researcher decides everything itself, using web/context7 research when available, and records the rationale.
argument_hint	<short project description> [--reset] [--auto]
runtime_intent	{"invokes_agent":"jdi-researcher"}
runtime_overrides	{"claude":{"allowed-tools":["Read","Write","Bash","Grep","Glob","AskUserQuestion","WebSearch","WebFetch","Agent"]},"copilot":{"tools":["read","write","grep","glob","terminal"]},"opencode":{"agent":"jdi-researcher","subtask":true},"antigravity":{"triggers":["/jdi-new","create project","new app"]}}

Initializes new JDI project. Runs research + key questions + generates PROJECT.md, ROADMAP.md, STATE.md, DECISIONS.md. - `description` (optional but recommended): short text of what to build. - `--auto` (alias: `--yolo`): fully autonomous — ZERO questions. Whatever the description does not answer, the researcher decides itself (researching via context7/web when available, else stack heuristics) and records the rationale. D-1 is auto-locked with a 1-line justification. The DoD baseline keeps the 5 derived candidates as-is. The richer the description, the fewer decisions are guessed — pair `--auto` with a dense prompt. `--auto` does NOT bypass the `--reset` confirmation (destructive stays human).

Examples:

• /jdi-new "TODO app .NET 10 + React 19"
• /jdi-new "Inventory REST API in Python + FastAPI" --auto
• /jdi-new "Go CLI tool for log parsing"
• /jdi-new (asker starts from scratch)

Step 1: Validation

test -d .jdi/ && {
  echo ".jdi/ already exists. Use /jdi-new --reset to start over (CAUTION: wipes state)."
  exit 1
}

# Suggest /jdi-adopt if directory is NOT empty (likely brownfield)
file_count=$(find . -maxdepth 3 -type f \
  -not -path './.git/*' -not -path './node_modules/*' \
  -not -path './.venv/*' -not -path './venv/*' \
  -not -path './target/*' -not -path './dist/*' -not -path './build/*' \
  -not -path './bin/*' -not -path './obj/*' \
  2>/dev/null | wc -l)

if [ "$file_count" -ge 3 ]; then
  # Ask before continuing — could be greenfield in monorepo
  echo "Directory has $file_count code files. Looks like existing project."
  echo "For brownfield projects, /jdi-adopt detects stack/code-design automatically."
  echo "Continue with /jdi-new anyway? (recommended: /jdi-adopt)"
  # AskUserQuestion: [Continue /jdi-new] / [Switch to /jdi-adopt] / [Cancel]
fi

PowerShell:

if (Test-Path .jdi) { Write-Error ".jdi/ already exists. Use /jdi-new --reset."; exit 1 }
$files = Get-ChildItem -Recurse -File -Depth 3 -ErrorAction SilentlyContinue |
  Where-Object { $_.FullName -notmatch '\\(\.git|node_modules|\.venv|venv|target|dist|build|bin|obj)\\' }
if ($files.Count -ge 3) {
  Write-Host "Directory has $($files.Count) files. Consider /jdi-adopt instead of /jdi-new."
  # Sequential AskUserQuestion
}

If --reset passed, AskUserQuestion confirms + wipes .jdi/.

Step 2: Spawn researcher

Invoke jdi-researcher passing the description AND auto=true when --auto (or its alias --yolo) was passed. Wait.

Step 3: Verify outputs

test -f .jdi/PROJECT.md || { echo "PROJECT.md not created"; exit 1; }
test -f .jdi/ROADMAP.md || { echo "ROADMAP.md not created"; exit 1; }
test -f .jdi/STATE.md || { echo "STATE.md not created"; exit 1; }
test -f .jdi/DECISIONS.md || { echo "DECISIONS.md not created"; exit 1; }
grep -q '## Definition of Done' .jdi/PROJECT.md || { echo "PROJECT.md missing § Definition of Done"; exit 1; }

Step 4: Create config.json (token/context budget)

If .jdi/config.json does not yet exist, write the default below. Defaults (200k context, 60/70% warn/critical, coverage 80%) cover 95% of cases. User edits if running a 1M-window model or wanting tighter thresholds.

{
  "$schema_version": "1.2",
  "context_window": 200000,
  "thresholds": {
    "warn_pct": 60,
    "critical_pct": 70
  },
  "budgets": {
    "max_context_chars": 6000,
    "max_plan_chars": 12000,
    "max_summary_chars": 8192
  },
  "compaction": {
    "archive_after": 5
  },
  "orchestration": {
    "mode": "standard",
    "source": "default"
  },
  "coverage_min": 80
}

Canonical reference for the default also lives in templates-jdi-folder/config.json (shipped by npm package) — for users wanting to regenerate manually.

Step 4b: Enhanced orchestration opt-in (host-neutral capability flag)

orchestration.mode tells later commands whether they MAY use optional multi-agent layers (advisory critics, parallel analysis), always degrading to the standard path when the host cannot fan out. Default standard keeps every command byte-identical on hosts without the capability — never assume it.

Determine the value once, here, in this top-level command turn — host capability signals are only visible now; sub-agents spawned by later commands cannot see them, so the choice MUST be persisted to the file:

1. Default: if this session is running under an enhanced / high-effort multi-agent orchestration mode, pre-select enhanced and set source: "detected". Otherwise standard.
2. Confirm (AskUserQuestion; fallback: numbered prompt — see <fallbacks> convention):

   Enable enhanced multi-agent orchestration when your assistant supports it?
   Adds opt-in advisory critics (e.g. a Definition-of-Done re-check at /jdi-verify).
   The standard single-agent path is used unchanged whenever it is unavailable.
   

   Options: [Enhanced — use extra agents when available] / [Standard — single-agent always (default)]
3. Write the chosen mode into the orchestration block; set source to "user" (or keep "detected" if the user accepts the detected default without changing it). Never store a token budget here — this is a boolean capability switch, not an accounting ledger.

Step 5: Confirm

{project_name} initialized. {N} phases planned in .jdi/.
Next: /jdi-bootstrap

- pre: directory without existing `.jdi/` (or `--reset`) - post: PROJECT.md + ROADMAP.md + STATE.md + DECISIONS.md + config.json created, initial commit made - `.jdi/` already exists -> suggest `--reset` or use current project - Researcher cancelled -> exit clean - Researcher failed -> show error, no commit