@hegemonart/get-design-done 1.57.2 → 1.58.0

package/dist/claude-code/.claude/skills/todo/SKILL.md

@@ -1,64 +0,0 @@
----
-name: gdd-todo
-description: "Design backlog - add/list/pick design tasks. Writes to .design/TODO.md."
-argument-hint: "<add|list|pick> [text]"
-tools: Read, Write, AskUserQuestion, mcp__gdd_state__get, mcp__gdd_state__add_decision, mcp__gdd_state__add_must_have
-disable-model-invocation: true
----
-
-# /gdd:todo
-
-**Role:** Design todo list. Three subcommands: `add`, `list`, `pick`. Backing store: `.design/TODO.md`. For items that are pipeline-level decisions or must-haves (not free-form backlog), route through the `gdd-state` MCP tools instead - see **Pipeline-linked items** below.
-
-## File format
-
-```markdown
-# Design Todos
-
-## P0 — Blocking
-- [ ] [2026-04-18] Fix contrast ratio on card headers
-
-## P1 — High
-- [ ] [2026-04-18] Align Button focus ring with tokens
-
-## P2 — Normal
-
-## P3 — Nice to have
-```
-
-Priority is encoded by section, not inline. Status markers: `[ ]` unchecked, `[x]` done, `[-]` `[in-progress]`, `[p]` `[promoted]`.
-
-## Subcommands
-
-### add [text]
-If text omitted, use `AskUserQuestion`: "What todo item? (include priority P0-P3 if known)". Parse leading `P[0-3]` token from text; default P2. Append under the right section as:
-```
-- [ ] [YYYY-MM-DD] <text without priority token>
-```
-Create TODO.md from the template above if missing.
-
-### list
-1. Call `mcp__gdd_state__get` → pipeline-level decisions + must-haves (shown as context at the top).
-2. Read `.design/TODO.md` (file outside the MCP catalog). Print all `- [ ]` and `- [-]` items grouped by priority section, with index numbers.
-
-### pick
-Read `.design/TODO.md`. Collect unchecked items. Use `AskUserQuestion` to let the user pick one. Rewrite the chosen line as:
-```
-- [-] [YYYY-MM-DD] [in-progress] <original text>
-```
-Print "Picked: <item>".
-
-## Pipeline-linked items
-
-When the user promotes a todo to a pipeline decision or must-have, route through MCP instead of TODO.md:
-
-- Decision → `mcp__gdd_state__add_decision` with `{ id: "D-XX", text: "...", status: "locked"|"tentative" }`.
-- Must-have → `mcp__gdd_state__add_must_have` with `{ id: "M-XX", text: "...", status: "pending" }`.
-
-## Constraints
-
-- Do not modify files outside `.design/`.
-- Preserve existing sections and ordering on write.
-- Do not mutate STATE.md directly - use the MCP tools above.
-
-## TODO COMPLETE

package/dist/claude-code/.claude/skills/turn-closeout/SKILL.md

@@ -1,95 +0,0 @@
----
-name: gdd-turn-closeout
-description: "Portable mirror of the gdd-turn-closeout Stop hook (D-11). Closes the events.jsonl gap at turn-end and surfaces a stage-completion or paused-mid-task nudge. Tail-called by orchestrator skills (/gdd:next, /gdd:design, /gdd:verify) at exit on the 13 non-Claude runtimes that lack a Stop hook surface. Idempotent, non-blocking, ≤10ms typical."
-argument-hint: "(none - reads .design/STATE.md and .design/telemetry/events.jsonl from cwd)"
-tools: Read, Bash
----
-
-# gdd-turn-closeout
-
-## Role
-
-You are a deterministic **closeout** skill. You close the per-turn telemetry gap on runtimes that don't expose a Stop event (codex, gemini, and 11 others). You are a code-level mirror of `hooks/gdd-turn-closeout.js` (D-10): same conditions, same idempotence, same emitted event shape. The only difference: the JS hook emits the nudge as `additionalContext` via the harness; this skill prints the nudge directly to the user. See `./../new-cycle/milestone-completeness-rubric.md` §"Task level" for the broader closeout discipline (what "turn complete" means within a stage).
-
-**When to invoke:** orchestrator skills (`/gdd:next`, `/gdd:design`, `/gdd:verify`) tail-call this skill as their final step before returning. Adoption is incremental - each orchestrator can wire the tail-call independently; the skill exists as a stable, callable surface today.
-
-## Invocation Contract
-
-- **Input**: none. Operates on `.design/STATE.md` and `.design/telemetry/events.jsonl` in cwd.
-- **Output**: at most one printed line - the nudge - or silent return.
-- **Latency budget**: ≤10ms typical. Read **only** STATE.md and the tail of events.jsonl.
-- **Idempotence**: if the most recent event is already a `turn_end` for the current `(stage, task_progress)` tuple, skip append but still print nudge.
-- **Non-blocking**: any I/O failure → silent return. This skill never gates the user.
-
-## Algorithm
-
-Execute in order; stop at the first early-return.
-
-### Step 1 - Read STATE.md
-
-Read `.design/STATE.md`. Missing or unreadable → **return silently** (no print, no append).
-
-### Step 2 - Parse `<position>` block
-
-Lightweight-parse only `<position>…</position>` (regex `/<position>([\s\S]*?)<\/position>/` then per-line `key: value`). Extract `stage`, `status`, `task_progress`. Do not invoke the full STATE parser.
-
-If `status != "in_progress"` → **return silently**. Pipeline is initialized / completed / blocked - no turn-end gap to close.
-
-### Step 3 - Tail the last event line
-
-Read **only the last 8 KiB** of `.design/telemetry/events.jsonl`. Treat as "stale by definition":
-
-- File missing.
-- File empty.
-- Last line fails JSON parse.
-- Last line's `timestamp` is missing or unparseable.
-
-Otherwise compute `now - last_event.timestamp`. Gap < 60 seconds → user is actively mid-turn → **return silently** (next real event closes the gap naturally). Bash one-liner for the tail: `tail -n 1 .design/telemetry/events.jsonl 2>/dev/null`.
-
-### Step 4 - Idempotence check, then append
-
-If last event is already `{type:"turn_end", stage:<same>, payload:{task_progress:<same>}}` for the exact `(stage, task_progress)` from Step 2: **skip append** but proceed to Step 5.
-
-Otherwise append one JSONL line to `.design/telemetry/events.jsonl`:
-
-```json
-{"type":"turn_end","timestamp":"<ISO 8601 now>","sessionId":"<session-id-or-'turn-closeout'>","stage":"<stage>","payload":{"task_progress":"<N/M>"},"_meta":{"source":"gdd-turn-closeout-skill"}}
-```
-
-Create `.design/telemetry/` if missing. Append is a single `appendFile`-equivalent call (writer assumes append-atomicity per Plan 20-06).
-
-### Step 5 - Print the nudge
-
-Match `task_progress` against `^(\d+)/(\d+)$`:
-
-- **Numerator equals denominator and denominator > 0** (e.g. `5/5`): `Stage <stage> complete — run /gdd:next or /gdd:reflect`.
-- **Otherwise** (mid-task, e.g. `3/7`, `0/0`, malformed): `Stage <stage> paused mid-task — resume with /gdd:resume`.
-
-One line exactly. No commentary - the nudge is the user-facing surface.
-
-## Failure Modes
-
-| Condition | Behavior |
-|-----------|----------|
-| STATE.md missing/unreadable | Silent return |
-| `<position>` absent or malformed | Silent return |
-| `status != "in_progress"` | Silent return |
-| `events.jsonl` missing | Treat as stale → fall through to append + nudge |
-| Last event unparseable | Treat as stale → fall through |
-| Last event < 60s old | Silent return |
-| Append fails (permission, disk full) | Print nudge anyway; do not surface I/O error |
-| Any uncaught throw | Silent return |
-
-## Equivalence with the JS hook
-
-This skill and `hooks/gdd-turn-closeout.js` MUST stay code-level equivalent: same four early-return branches, same 60-second staleness threshold, same idempotence guard (`type=turn_end, stage, payload.task_progress`), same event shape (only `_meta.source` differs: `gdd-turn-closeout` vs `gdd-turn-closeout-skill` so reflector telemetry can distinguish hook-driven vs skill-driven). Same nudge wording for both N/N and mid-task cases. Change one → change the other in the same plan. Plan 25-09's `tests/turn-closeout-hook.test.cjs` covers the JS hook.
-
-## Non-Goals
-
-- **Not a state writer.** Never edits STATE.md. The events.jsonl append is the only side effect.
-- **Not a stage transition.** `turn_end` is within-stage, not a state-machine move; downstream tools ignore it for transition gating.
-- **Not a Stop-event harness.** Cross-runtime Stop-event support at the harness level is out-of-scope for Phase 25 (CONTEXT.md OOS).
-
-## Integration Point
-
-Canonical tail-call sites per D-11: `/gdd:next`, `/gdd:design`, `/gdd:verify`. Each orchestrator's final step before returning to the user invokes `gdd-turn-closeout`. Tail-call wiring intentionally not part of v1.25 (Plan 25-04 ships only the callable surface).

package/dist/claude-code/.claude/skills/undo/SKILL.md

@@ -1,31 +0,0 @@
----
-name: gdd-undo
-description: "Safe design change revert. Uses git log to find design commits, checks dependencies, reverts safely."
-argument-hint: "[<commit SHA>]"
-tools: Read, Write, Bash, AskUserQuestion
-disable-model-invocation: true
----
-
-# /gdd:undo
-
-Safe rollback for design-related commits. Uses git history plus file-overlap dependency checks before reverting.
-
-## Steps
-
-1. **List candidates**: Run `git log --oneline -n 20 --grep="design\|feat\|fix"` via Bash. Filter for commits whose scope matches phase prefixes (`design(`, `feat(07`, etc.) or touch files under `src/` related to design changes.
-2. **Present choice**: If no SHA was passed as argument, print the last 10 design-related commits and ask (AskUserQuestion): "Which commit to undo? (or enter commit SHA)"
-3. **Dependency scan**: For the chosen SHA, run `git show --name-only <sha>` to get touched files. Then run `git log <sha>..HEAD --name-only` to find later commits that modified any of those same files. These are potential dependencies.
-4. **Dependency decision**:
-   - If none found: proceed to step 5.
-   - If found: print "Reverting this commit may break: <later commits>. Proceed anyway? (yes/no)" and abort on "no".
-5. **Stage the revert**: Run `git revert <sha> --no-commit`. Show the resulting diff via `git diff --cached`.
-6. **Confirm**: Ask (AskUserQuestion): "Apply this revert? (yes/no)"
-7. **Commit or abort**: On yes, `git commit -m "revert: <original subject>"`. On no, `git reset` to unstage.
-
-## Do Not
-
-- Do not force-push.
-- Do not revert commits on `main` without the user confirming branch context.
-- Do not silently skip dependency warnings.
-
-## UNDO COMPLETE

package/dist/claude-code/.claude/skills/unlock-decision/SKILL.md

@@ -1,54 +0,0 @@
----
-name: gdd-unlock-decision
-description: "Reopens a LOCKED design decision - the only escape hatch from the hard lock. Requires an explicit --approver and writes an audit entry, then moves the decision locked → reviewing (via scripts/lib/collab/review-queue.cjs). Previews the audit record before writing; never unlocks silently. Use when a locked decision genuinely must change (a later constraint invalidated it) and a reviewer has signed off."
-argument-hint: "<decision-id> --approver <who> [--reason <text>] [--dry-run]"
-user-invocable: true
-tools: Read, Write, Bash, Grep, Glob
----
-
-# /gdd:unlock-decision
-
-A `locked` decision is hard - it cannot be amended. This skill is the **only** way back, and it is
-deliberately heavyweight: it requires a named approver and records an audit entry, so reopening a
-locked decision is always traceable. Contract: `../../reference/multi-author-model.md`.
-
-## Invocation
-
-| Command | Behavior |
-|---|---|
-| `/gdd:unlock-decision <id> --approver <who>` | Unlock `<id>` (locked → reviewing), recording the approver. |
-| `/gdd:unlock-decision <id> --approver <who> --reason <text>` | Same, with a reason in the audit entry. |
-| `/gdd:unlock-decision <id> --dry-run` | Preview the audit record + the resulting state; change nothing. |
-
-## Steps
-
-1. **Validate args.** `<decision-id>` required; `--approver` required (a non-empty name). Missing
-   `--approver` → print the usage + exit without changing anything.
-2. **Load the entry** from `.design/reviews/<decision-id>/state.json`. Not found → report it. Not in
-   `locked` state → print `unlock-decision: <id> is not locked (state: <state>); nothing to unlock.`
-3. **Preview.** Show the audit entry that will be appended (`{ action: unlock, from: locked, to:
-   reviewing, approver, reason }`) and the resulting state. If ``--dry-run``, stop here.
-4. **Apply** via the pure helper:
-
-   ```bash
-   node -e '
-     const fs = require("fs");
-     const rq = require("./scripts/lib/collab/review-queue.cjs");
-     const p = process.argv[1], approver = process.argv[2], reason = process.argv[3] || "";
-     const entry = JSON.parse(fs.readFileSync(p, "utf8"));
-     const next = rq.unlock(entry, { approver, reason });
-     fs.writeFileSync(p, JSON.stringify(next, null, 2) + "\n");
-     console.log(JSON.stringify(next.audit[next.audit.length - 1]));
-   ' ".design/reviews/<id>/state.json" "<who>" "<reason>"
-   ```
-
-5. **Report** the new state + the recorded approver. The decision is now `reviewing` and amendable
-   again; it must be re-reviewed + re-locked to become durable.
-
-## Output
-
-End with:
-
-```
-## UNLOCK-DECISION COMPLETE
-```

package/dist/claude-code/.claude/skills/unpin/SKILL.md

@@ -1,31 +0,0 @@
----
-name: gdd-unpin
-description: "Removes pinned skill aliases across harness dirs, deleting only stubs that carry the gdd pin marker. Use when you no longer want a pinned shortcut and want hand-written skills left untouched."
-argument-hint: "<skill-name>"
-tools: Read, Bash
----
-
-# /gdd:unpin
-
-**Role:** Remove the pinned shortcut aliases for one gdd skill from every installed harness `skills/` directory. Only stubs that carry the marker line `<!-- gdd-pinned-skill source=<skill> -->` as their first non-empty line are deleted, so a hand-authored or unrelated SKILL.md is never touched.
-
-## Steps
-
-1. **Read the argument.** The skill name to unpin comes from `$ARGUMENTS` (for example `darkmode`). If it is empty, ask the user which skill to unpin and stop.
-
-2. **Run the unpin CLI.** Invoke the shipped script, passing the skill name. The plugin root resolves via `CLAUDE_PLUGIN_ROOT` (falling back to the current directory when that variable is absent):
-
-    ```bash
-    node "${CLAUDE_PLUGIN_ROOT:-$(pwd)}/scripts/lib/pin/cli.cjs" unpin "<skill-name>"
-    ```
-
-    The CLI scans every harness `skills/` directory under the current project, checks each candidate stub for the gdd pin marker, deletes the ones that carry it, and refuses (skips with a warning) any file that does not.
-
-3. **Report the result.** The CLI prints one line per harness it removed a stub from, plus a warning for any file it refused to delete. Relay that summary verbatim. Exit codes: 0 means at least one stub was removed, 1 means nothing was removed (no matching pinned stubs found), 2 means an error.
-
-## Do Not
-
-- Do not delete skill files by hand. Always go through the CLI so the marker check protects hand-written skills.
-- Do not force-remove a file the CLI refused. A refusal means the file lacks the gdd pin marker and is not a pinned alias; leave it in place.
-
-## UNPIN COMPLETE

package/dist/claude-code/.claude/skills/update/SKILL.md

@@ -1,56 +0,0 @@
----
-name: gdd-update
-description: "Update get-design-done to the latest release. Preserves .design/config.json and ./.claude/skills/."
-argument-hint: "[--dry-run] [--version <tag>] [--show-privacy-diff]"
-tools: Read, Write, Bash
-disable-model-invocation: true
----
-
-# gdd-update
-
-Updates the `get-design-done` plugin to the latest release (or a specific tag), preserving user-local state.
-
-## Steps
-
-1. **Pre-flight check** - read current version from `.claude-plugin/plugin.json` (fallback: `plugin.json` at project root). Print current version and, when available, the latest release tag from GitHub.
-2. **Dry-run** - if `--dry-run` is passed, print the planned steps and exit without making changes.
-3. **Backup** - read `.design/config.json` into memory. Snapshot the file list under `./.claude/skills/` so we can detect collisions later. **Additionally, snapshot the text contents of `scripts/lib/pseudonymize.cjs`, `scripts/lib/issue-reporter/payload-assembly.cjs`, and `scripts/lib/issue-reporter/destination.cjs` to a tempdir (mirroring the same relative paths under that tempdir's root). This is the OLD tree used by the privacy-diff step below. Also read the previous-version string from `.design/privacy-diff-last-version.txt` if it exists, into a variable `prevVersion` (null if the file is absent).** (Plan 30-07 D-09)
-4. **Pull update** - run `claude plugin install hegemonart/get-design-done` (or `claude plugin install hegemonart/get-design-done@<tag>` when `--version <tag>` is passed). This re-syncs all plugin files from the release.
-5. **Restore config** - write `.design/config.json` back from the in-memory backup. The installer may reset the config to defaults; this step guarantees user settings survive.
-5.5. **Privacy diff** - `const pd = require('./scripts/lib/issue-reporter/privacy-diff.cjs');` Compute `showDiff = pd.shouldAutoShow(prevVersion, currentVersion, oldTreeTempdir, repoRoot) || flags.showPrivacyDiff`. If `showDiff` is true: call `pd.computePrivacyDiff(oldTreeTempdir, repoRoot)`, pass the result to `pd.renderPrivacyDiff`, and print the returned markdown to stdout under a clear "## Privacy-critical changes" banner. If `--show-privacy-diff` was passed but `prevVersion === null` (first-ever upgrade), print: "Privacy diff requested but no previous version is recorded. Snapshot file `.design/privacy-diff-last-version.txt` will be written now; the next upgrade will be able to diff against this version." After printing (or skipping), write `currentVersion` to `.design/privacy-diff-last-version.txt` so the next upgrade compares against THIS version. (Plan 30-07 D-09)
-6. **Preserve local skills** - `./.claude/skills/` is project-local and outside the plugin tree. Re-list the directory and confirm none of the pre-update files disappeared. Warn loudly if any did.
-7. **Post-update advisory** - print:
-
-   > Run `/gdd:reapply-patches` if you have customized any `reference/` files to restore your modifications.
-
-7.5. **Deprecation advisory** (Phase 39.5) - load the path-migration registry and report anything that
-   crossed into `deprecated` or `removed` over the `[prevVersion → currentVersion]` window:
-
-   ```bash
-   node -e '
-     const fs = require("fs");
-     const dr = require("./scripts/lib/deprecation-registry.cjs");
-     const entries = dr.parseDeprecations(fs.readFileSync("reference/DEPRECATIONS.md","utf8"));
-     const crossed = entries.filter(e =>
-       dr.classify(e, currentVersion) !== "pending" &&
-       (prevVersion == null || dr.classify(e, prevVersion) !== dr.classify(e, currentVersion)));
-     console.log(JSON.stringify(crossed));
-   '
-   ```
-
-   If any entry crossed, print a `## Deprecations in this update` list (old → new + status) and point
-   the user at **`/gdd:migrate`** to rewrite their local references. If none crossed, say nothing.
-
-8. Print the new version and the changelog URL (`https://github.com/hegemonart/get-design-done/releases`).
-
-## Output
-
-End every invocation with:
-
-```
-## UPDATE COMPLETE
-```
-
-## Implementation note
-
-The actual update mechanism is the standard `claude plugin install` re-install path. This skill only orchestrates the pre-/post-preservation steps around it.

package/dist/claude-code/.claude/skills/using-gdd/SKILL.md

@@ -1,78 +0,0 @@
----
-name: using-gdd
-description: "Use when starting any GDD session - establishes how to find and apply GDD skills."
-disable-model-invocation: true
----
-<SUBAGENT-STOP>
-
-# Using GDD
-
-This is the bootstrap discipline contract for every Get Design Done session. Read it
-first; it tells you how to find and apply the right GDD skill before you act.
-
-## The 1% rule
-
-**If you think there is even a 1% chance a skill might apply, you ABSOLUTELY MUST invoke the skill.**
-
-In GDD, almost every request maps to a pipeline stage - brief, explore, plan, design,
-verify - or to a cross-cutting skill (discuss, audit, style, darkmode). When in doubt,
-search for and read the skill's body. The cost of reading a skill is trivial; the cost of
-free-handing a stage is rework, scope creep, and a broken pipeline state.
-
-## Red flags - Thought → Reality
-
-When you catch yourself thinking any of the following, STOP and check for a skill.
-
-| Thought | Reality |
-| --- | --- |
-| This is just a simple design question. | Questions are tasks. Check for a skill. |
-| I'll just tweak the CSS directly. | Token changes go through the pipeline - check /gdd:design. |
-| I already know the codebase, skip explore. | Explore probes connections you haven't re-checked this cycle. |
-| This change is too small to plan. | Plan-skipped tasks blow scope per cycle telemetry. Run /gdd:plan. |
-| I can write the brief later. | No brief means no shared problem statement - /gdd:brief comes first. |
-| The user clearly wants X, I'll skip discuss. | Ambiguity hides here. /gdd:discuss surfaces the real constraint. |
-| I'll verify by eyeballing it. | Verification is a stage with criteria - run /gdd:verify, don't guess. |
-| It's obviously a dark-mode tweak. | Color-scheme work has its own skill - check /gdd:darkmode. |
-| Let me just compare these two designs quickly. | Comparison is an audit task - /gdd:compare has the rubric. |
-| This is a one-off, no skill needed. | "One-off" is the most common rationalization in the telemetry. Check anyway. |
-| I'll refactor the style tokens by hand. | /gdd:style owns token edits so the pipeline stays consistent. |
-| The audit can wait until after I ship. | An un-audited cycle is an unverified cycle - /gdd:audit before close. |
-
-## Skill priority order
-
-When more than one skill could apply, resolve in this order:
-
-1. **Process** - brief / explore / discuss. Establish the problem and context first.
-2. **Implementation** - design / style / darkmode. Only after process is settled.
-3. **Audit** - verify / compare / audit. Close the loop before declaring done.
-
-Never run an Implementation skill before the Process skills that gate it have produced
-their artifact. Never declare a cycle complete without an Audit skill.
-
-## Instruction priority
-
-When instructions conflict, obey this precedence (highest first):
-
-1. **The user's CLAUDE.md** - project- and user-level directives always win.
-2. **GDD skills** - the skill body is the source of truth for how a stage runs.
-3. **Defaults** - your own general training and habits come last.
-
-If a GDD skill contradicts the user's CLAUDE.md, the CLAUDE.md wins and you flag the
-conflict. If your instinct contradicts a GDD skill, the skill wins.
-
-## GDD pipeline flow
-
-The core flow is **Brief → Explore → Plan → Design → Verify**, with branch points:
-
-- **Brief** captures the problem (`/gdd:brief`). Branch: a rough idea can sketch or spike
-  off the brief before exploration; a changed problem loops back via `--re-brief`.
-- **Explore** scans the codebase and connections (`/gdd:explore`) - even on a familiar
-  repo, because connections drift each cycle.
-- **Plan** decomposes work into tasks (`/gdd:plan`). Skipping it is the top cause of scope
-  blow-up; small tasks still get a plan.
-- **Design** implements (`/gdd:design`, with `/gdd:style` and `/gdd:darkmode` as
-  implementation peers). Implementation never runs ahead of an approved plan.
-- **Verify** checks against criteria (`/gdd:verify`), then `/gdd:audit` / `/gdd:compare`
-  close the loop. On pass the cycle completes; on fail it loops back to the failing stage.
-
-`/gdd:discuss` runs alongside any stage to resolve ambiguity before it propagates.

package/dist/claude-code/.claude/skills/verify/SKILL.md

@@ -1,113 +0,0 @@
----
-name: verify
-description: "Stage 5 of 5 orchestrator that spawns design-auditor, design-verifier, and design-integration-checker in sequence, interprets pass/gap result, and drives the gap-response loop (inline fix, save-and-exit, or accept-as-is). Use when implementation is complete and ready for final pre-ship verification. Activates for requests involving checking finished UI against the design system, running a pre-ship review, or final verification."
-argument-hint: "[--auto] [--post-handoff]"
-user-invocable: true
-tools: mcp__gdd_state__get, mcp__gdd_state__transition_stage, mcp__gdd_state__add_must_have, mcp__gdd_state__add_blocker, mcp__gdd_state__resolve_blocker, mcp__gdd_state__update_progress, mcp__gdd_state__set_status, mcp__gdd_state__checkpoint, mcp__gdd_state__probe_connections
----
-
-# Get Design Done - Verify
-
-**Stage 5 of 5** in the get-design-done pipeline. Thin orchestrator. Verification intelligence lives in three agents: design-auditor, design-verifier, and design-integration-checker.
-
-Full procedure detail: `./verify-procedure.md`.
-
----
-
-## State Integration
-
-1. `mcp__gdd_state__transition_stage` with `to: "verify"`; on gate failure surface `error.context.blockers` to the user without advancing.
-2. `mcp__gdd_state__get` -> snapshot `state`. Read `state.must_haves` (verification checklist - each M-XX flips to `pass` or `fail`).
-3. **Quality-gate gate (D-08, D-09)** - inspect `state.quality_gate?.run?.status`:
-   - `"fail"` -> refuse to advance; call `mcp__gdd_state__add_blocker` with the iteration count + `commands_run`; exit. Do NOT open the stage.
-   - `"timeout"` / `"skipped"` -> print one-line warning naming the status + `commands_run`, continue normally (signals, not walls).
-   - `"pass"` / `null` -> continue silently.
-   Full decision tree: `./verify-procedure.md` §Quality-gate gate.
-4. Resume detection - if `state.position.status==in_progress` and `.design/DESIGN-VERIFICATION.md` exists: RESUME to Step 2 (gap-response loop). Otherwise call `mcp__gdd_state__update_progress` with `task_progress: "0/3"`, `status: "in_progress"` and proceed.
-5. Missing STATE.md is a hard block - verify is never the entry point; upstream stages own bootstrap.
-
-**Flipping a must-have status:** `mcp__gdd_state__add_must_have` with the SAME `id` updates in-place (no separate update tool). Detail: `./verify-procedure.md` §Flipping a must-have status.
-
----
-
-## Connection probes
-
-Run preview / storybook / chromatic probes at stage entry, then issue ONE batched `mcp__gdd_state__probe_connections` call with all results. Full probe specs (project detection, dev-server probe, CLI presence, token check) and downstream loops (storybook a11y, chromatic visual delta) are in `./verify-procedure.md` §Connection Probes.
-
----
-
-## Prerequisites + flags
-
-- **DESIGN-PLAN.md prerequisite** (normal mode): missing -> block with "Verify requires DESIGN-PLAN.md. Run `/gdd:plan` first, or use `--post-handoff` if starting from a Claude Design handoff bundle."
-- **Post-handoff mode** (`--post-handoff` OR STATE.md `status: handoff-sourced`): skip the DESIGN-PLAN.md check; pass `post_handoff: true` + `handoff_path` to design-verifier; DESIGN-VERIFICATION.md gains a `## Handoff Faithfulness` section. Detail: `./verify-procedure.md` §Post-Handoff Mode.
-- **Flags:** `--auto` -> `auto_mode=true` (no interactive prompts; on gaps: save-and-exit); `--post-handoff` -> see above.
-- **Parallelism decision:** read `.design/config.json` + `reference/parallelism-rules.md`. Default serial (verifier depends on auditor output). Record via `mcp__gdd_state__set_status` before spawning. Detail: `./verify-procedure.md` §Parallelism Decision.
-
----
-
-## Step 1 - Spawn Auditor + Verifier + Integration Checker
-
-Initialize the fix-loop iteration counter to 0. Each full checker is preceded by a cheap Haiku gate that may return `{spawn: false}` to short-circuit (lazy-gate pattern from Plan 10.1-04 / D-21); skipped agents append `lazy_skipped: true` to `.design/telemetry/costs.jsonl`.
-
-**1a. design-auditor** (retrospective 7-pillar audit) -> `.design/DESIGN-AUDIT.md`. Wait for `## AUDIT COMPLETE`, then `mcp__gdd_state__update_progress` `task_progress: "1/3"`.
-
-**1b-gate -> 1b. design-verifier** (5-phase verification, reads auditor output) -> `.design/DESIGN-VERIFICATION.md`. Wait for `## VERIFICATION COMPLETE`, then `task_progress: "2/3"`.
-
-**1c-gate -> 1c. design-integration-checker** (per-decision wiring check on each D-XX in DESIGN-CONTEXT.md) -> reports Connected / Orphaned / Missing counts. Wait for `## INTEGRATION CHECK COMPLETE`, then `task_progress: "3/3"`.
-
-Full agent prompts, lazy-gate decision logic, and telemetry-row shapes: `./verify-procedure.md` §Step 1.
-
----
-
-## Step 2 - Interpret Result
-
-Consolidate gaps from both sources: verifier `## GAPS FOUND` (G-NN entries) and integration-checker (Orphaned -> MAJOR, Missing -> BLOCKER).
-
-- **No gaps (PASS)** -> for each M-XX in the entry snapshot: `add_must_have` with `status: "pass"`. Proceed to Stage exit.
-- **Gaps + `auto_mode=true`** -> preserve DESIGN-VERIFICATION.md, `set_status: "blocked"`, `add_blocker` with the gap count, exit with the failure message.
-- **Gaps + `auto_mode=false`** -> proceed to Step 3.
-
-Detail: `./verify-procedure.md` §Step 2.
-
----
-
-## Step 3 - Gap Response Loop
-
-Present the gap summary + 3-option menu (`[1] Fix now`, `[2] Save and exit`, `[3] Accept as-is`).
-
-- **[1] Fix now** -> if iteration counter >= 3 fall back to [2]; otherwise increment counter, spawn `design-fixer` for BLOCKER+MAJOR gaps (each fix is an atomic `fix(design-gap-GNN):` commit), wait for `## FIX COMPLETE`, then re-spawn `design-verifier` with `re_verify=true` and loop to Step 2.
-- **[2] Save and exit** -> preserve DESIGN-VERIFICATION.md, `set_status: "blocked"`, `add_blocker`, `checkpoint`, exit.
-- **[3] Accept as-is** -> flip each unmet M-XX to `status: "fail"`, `add_blocker` with "accepted with N unresolved gaps", proceed to Stage exit.
-
-Full prompts + branching: `./verify-procedure.md` §Step 3.
-
----
-
-## Stage exit
-
-1. `mcp__gdd_state__update_progress` -> `task_progress: "<verified>/<total>"`, `status: "verify_complete"`.
-2. `mcp__gdd_state__set_status` -> `"pipeline_complete"` (all pass, no gaps) or `"verify_failed_requires_loop"` (gaps remain).
-3. `mcp__gdd_state__checkpoint` - stamps `last_checkpoint` and appends `verify_completed_at`. No direct STATE.md writes.
-
-## After Completion
-
-Print the `=== Verify complete ===` summary (status, gap counts, agent paths, next-step suggestion) from `./verify-procedure.md` §After Completion.
-
-<HARD-GATE>
-Do NOT mark the cycle complete until the user has reviewed `.design/DESIGN-VERIFICATION.md`. If this project uses a custom `.design` location, read the artifact path from `.design/STATE.md` rather than assuming the default.
-</HARD-GATE>
-
-## Rationalizations - Thought to Reality
-
-The reasons an agent gives to skip or weaken verification, and what each one lets through:
-
-| Thought | Reality |
-|---------|---------|
-| "The implementation looks right, I can skip the verifier spawn." | Skipping verification ships unchecked must-haves; the 5-phase verifier exists to catch what "looks right" misses. |
-| "The integration-checker is redundant with the auditor." | The auditor scores quality; the integration-checker proves each D-XX is actually wired - an orphaned decision passes the audit but fails the product. |
-| "These gaps are minor, I'll accept-as-is without a blocker." | Accept-as-is without recording the unresolved gaps erases the trail; the next cycle re-discovers them from scratch. |
-| "The quality gate timed out, I'll treat that as a pass." | Timeout is a signal, not a pass - masking it lets a genuinely failing gate slip through to ship. |
-| "I'll loop the fixer a fourth time to clear the last gap." | The 3-iteration cap exists because a gap surviving three fixes is a design problem, not a code problem - save and escalate. |
-| "Post-handoff bundles don't need the faithfulness check." | Skipping the handoff-faithfulness section means a divergence from the source design ships unflagged. |
-
-## VERIFY COMPLETE