Help — what next?

Language, command namespaces (/web-studio:<name> in plugin mode), the built-in /code-review collision and the closing question: docs/coordination-rules.md § Skill conventions.

Read-only: no file writes; the only commands are diff -rq and claude plugin list --json (Phase 4 version drift) — the frontmatter pre-grants exactly those two. A quick orientation, not a full audit (that is /adopt).

Names:

• <plugin root> — the path on the "Plugin root:" line of the session-start context, printed only in plugin mode. A /cleared session may lack the line: then installPath from claude plugin list --json; no web-studio entry there either → copy mode.
• "A project that has code" — a manifest (go.mod, composer.json, package.json, …) or source files outside .claude/, docs/ and production/.
• Every command named in the answer comes from the catalog or from a skill's own frontmatter, never from memory. A sub-command only when the skill's argument-hint lists it: there is no /dev-story complete S-NNN, a story is closed by /story-done; an invented shape sends the user to a dead end with the studio's authority behind it.

Phase 0: Reference modes

Argument commands or guide [topic] → read references/reference-modes.md and follow it, then stop: verdict READY, a text line, no closing AskUserQuestion (rule 7: nothing to decide).

Phase 1: Catalog and reads

1. Read .claude/docs/workflow-catalog.yaml: phases, steps, artifact.glob. Missing → the studio is not initialised: answer "run /init" and stop.
2. technical-preferences.md whose **Type** field is still [TO BE CONFIGURED], on a project that has code → initialised but not adopted: NEXT is /adopt full. Check the field itself, never a grep of the whole file: the template's header comment names the placeholder too.
3. Reads of its own (Read/Glob; Phases 2–4 reuse them, nothing is read twice):
   • production/stage.txt and production/review-mode.txt;
   • the updated: line of .claude/docs/stack-reference/index.md (absent → ?);
   • the newest docs/adoption-plan-*.md by name (none → none);
   • the name: of every .claude/skills/*/SKILL.md and <plugin root>/skills/*/SKILL.md;
   • when production/roadmap.md exists, its first open - [ ] line and the count of open lines: a NEXT candidate (rank 7), always printed as Roadmap: N open — first: <line>.

Phase 2: Where we are

NEXT precedence, highest first (each step cites its rank):

1. Overrides — open BLOCKING findings without a story, the game gate (step 6).
2. The user's argument "just finished X" (step 4).
3. Open items of a COMPLIANT adoption plan (step 5).
4. Unmet required steps of earlier phases (step 3).
5. Steps of the current phase (step 2).
6. Open items of any other adoption plan (step 5).
7. The roadmap's first open line (step 7).

Steps:

1. Stage from production/stage.txt; otherwise infer it from artefacts (the first phase with an unmet required step).
2. Steps of the current phase (rank 5), each checked by its glob: ✅ done / ⬜ missing / 🔁 repeatable.
3. Earlier phases (rank 4). A brownfield project that started at build/operate still owes the required steps of every earlier phase (check their globs); an unmet one (e.g. no docs/architecture/test-strategy.md) precedes the current phase's steps.
4. The user's argument (rank 2, "just finished X"): find step X in the catalog; NEXT is the next step of its phase (or the first step of next_phase). "finished security-audit" → /dependency-audit//harden in hardening.
5. Adoption plan (rank 3 or 6). Count the open items (- [ ]) of the Phase 1 plan.
   • A plan written before 0.7.0 has a numbered table, no checkboxes: its rows are the items, said in one line (plan in table format — /adopt full rewrites it as checkboxes). Never report 0 open for such a plan.
   • Show Adoption plan: N open.
   • Verdict COMPLIANT → rank 3: its open items come before unmet required steps of earlier phases, which are shown as ⬜ (not migrated by decision — see adoption plan) and are not NEXT.
   • Otherwise rank 6: when the phase has no unmet required step (typical for operate), the first open plan item is NEXT.
   • Never call the NEXT you name "low-value"; if the plan ranks it low, name the plan's own first item.
6. NEXT overrides (rank 1).
   • production/findings.md has open BLOCKING findings without a story → NEXT is /create-stories.
   • Game project (technical-preferences type game / game+backend): every story of the first feature Done and production/releases/gate-prototype.md missing → NEXT is /game-concept gate.
7. Roadmap (rank 7). When ranks 1–6 claim nothing, the roadmap's first open line (Phase 1) is NEXT: /dev-story S-NNN for a story line, the line's own command otherwise (a ⛔ [D-NN] marker → /backlog review/the decision named, never the blocked story).

Phase 3: Uncatalogued skills

1. Compare the skill names from Phase 1 with the catalog's command:. Show up to 8 relevant to the phase as "Also available".
2. Shadow copies: plugin installed and .claude/skills/ holds skills of the same names → the copies win for a bare command, /web-studio:<name> runs the plugin. Print one line ("N project copies shadow the plugin: /update to check and remove them"); the project's own skills are named separately, they stay.

Phase 4: Output

Stage: [label] ([N/M] required done)
✅ /setup-stack — stack pinned
⬜ /product-spec — no docs/specs/product-spec.md   ← NEXT
🔁 /feature-spec — 2 specs exist
Adoption plan: 3 open — first: /threat-model (docs/adoption-plan-2026-09-08.md #2)
Roadmap: 12 open — first: S-012 · Repository layer (production/roadmap.md)
Next: /product-spec  (why: nothing to check features against without it)
Also available: /stack-update, /team-feature …
Docs: /help commands (every command) · /help guide (what to run in every situation) · /help guide 10.7 (one section)

The Docs: line is always printed: it is how users discover the reference modes.

Add one line each, when it applies — read references/signals.md for the exact line and command:

1. Stack reference older than 60 days → /stack-update.
2. Version drift — evidence, never a stamp. Plugin mode: diff -rq <plugin root>/docs .claude/docs (and rules), claude plugin list --json (session vs installed). Copy mode: not checked here — nothing in the session names the kit path; /update --kit <path> compares the .claude/.web-studio-version stamp with the kit's plugin.json. Plugin-mode drift → /update, also a closing option.
3. Session state: production/session-state/active.md exists → its Task:/Next:.
4. Backlog: open ideas in production/backlog.md, oldest past 30 days or last-review older than 7 days → /backlog review; never a closing option.
5. Findings: open BLOCKING findings without a story → /create-stories (rank 1 made it NEXT).
6. External signals: a red CI, a failed deploy, a billing or access problem seen in session-state, a tech-debt CRITICAL → one Attention: line each with the fixing command or place; never investigated (no gh run, no logs: orientation, not diagnosis), never a closing option.
7. Sprint over and not closed: the latest production/sprints/sprint-NN.md, header only → Attention: … /retrospective NN, rules of 6.
8. Deploy artefacts: build phase, a Deploy target in technical-preferences, no docs/ops/deploy.md → the "Deploy artefacts" story is missing, /create-stories adds it.
9. Session-start signals: a merged branch, a CLAUDE.md template placeholder, a stack reference older than 60 days or an open Gate: printed at session start → repeated here, one line each.

Verdict: READY.

Next step — one AskUserQuestion about the pipeline only: the "Next" command (Recommended) · up to two "Also available" commands relevant to the phase · nothing now. Attention: items are never options.

The answer is not executed here: help runs on Haiku and a skill started from its answer inherits the turn's model. After the answer print one line, Run: /web-studio:<command> <args> (copy mode /<command>), and end the turn; the user sends it next.