harnessed 2.0.1 → 3.0.1

package/workflows/task/code/workflow.yaml

@@ -0,0 +1,55 @@
+# workflows/task/code/workflow.yaml — Phase v3.0-3.4 W0 T3.4.W0.7 (D-09 + D-05 + D-15)
+#
+# Stage ③.b 子任务编码 — karpathy 4 心法 always-on (Think Before Coding / Simplicity First /
+# Surgical Changes / Goal-Driven Execution) + mattpocock conditional route (zoom-out 陌生 /
+# improve-codebase-architecture 周期审查 / diagnose bug root cause unknown)。
+# 02-progress: planning-with-files Claude Code plugin /plan slash cmd 更新 progress.md
+# (per D-15 + Q-AUDIT-5a Option A — plugin 真接, NOT npm SDK NOT fs.writeFile self-impl)。
+#
+# Sister refs:
+#   - .planning/phase-v3.0-3.2/RESEARCH-workflows.md § Area 2 task/code sub yaml example
+#   - ~/.claude/CLAUDE.md "Execute 阶段" 节 karpathy 4 心法 + mattpocock 招式按需召唤
+#   - workflows/capabilities.yaml zoom-out + improve-codebase-architecture + diagnose +
+#     planning-with-files (Phase 2.3 W0.1 SHIPPED)
+#   - workflows/disciplines/karpathy.yaml (Phase 3.3 W0.4 SHIPPED, L0 substrate always-on)
+#   - src/workflow/schema/workflow.ts WorkflowSchemaV3 (Phase 3.3 W0.5 SHIPPED)
+#
+# disciplines_applied: 6 default 全开 per D-09 L0 Discipline Substrate always-on
+# tools_available: 4 entry ∈ capabilities.yaml entry set (T3.3.W0.10 Contract C1 catch)
+
+schema_version: harnessed.workflow.v3
+workflow: task-code
+description: |
+  Stage ③.b 子任务编码 — karpathy 4 心法 always-on + mattpocock conditional route
+  (zoom-out 陌生模块 / improve-codebase-architecture 周期审查 / diagnose bug root cause
+  unknown)。02-progress: planning-with-files Claude Code plugin slash cmd `/plan`
+  更新 progress.md 跟踪 subtask 进度 (per D-15 + Q-AUDIT-5a Option A LOCKED)。
+
+disciplines_applied: [karpathy, output-style, language, operational, priority, protocols]
+tools_available:
+  - zoom-out
+  - improve-codebase-architecture
+  - diagnose
+  - planning-with-files
+
+phases:
+  - id: 01-code
+    name: code (karpathy 心法 always-on + mattpocock conditional route)
+    upstream: karpathy
+    model: sonnet
+    max_iterations: '{{ defaults.ralph_max_iterations.task-code.01-code }}'
+    invokes_tools:
+      - if: 'phase.unfamiliar_module == true'
+        tool: zoom-out
+      - if: 'phase.architecture_health_audit == true'
+        tool: improve-codebase-architecture
+      - if: 'subtask.bug_root_cause_unknown == true'
+        tool: diagnose
+
+  - id: 02-progress
+    name: progress (planning-with-files /plan — update progress.md)
+    upstream: planning-with-files
+    capability: '{{ capabilities.planning-with-files.cmd }}'
+    invokes: '/plan'
+    model: haiku
+    artifacts_expected: [progress.md]

package/workflows/task/deliver/SKILL.md

@@ -0,0 +1,118 @@
+---
+name: task-deliver
+description: |
+  task-deliver workflow v3 — Stage ③.d 子任务交付 sub-workflow (ralph-loop COMPLETE
+  wrapper + Agent Teams conditional escalation + R20.10 explicit max_iterations_exceeded
+  fallback)。2-phase composition: 01-deliver (ralph-loop SDK wrapper with completion_promise
+  verbatim "COMPLETE" + parallelism judgments.parallelism-gate.ralph-loop-wrapper.fires +
+  fallback emit_warning_and_halt exit_code 1) → 02-progress-mark (Claude Code plugin
+  /plan mark subtask complete in progress.md)。
+  schema_version: harnessed.workflow.v3 with disciplines_applied [6] + tools_available
+  [ralph-loop, agent-teams-create, agent-teams-send-message, agent-teams-shutdown,
+  planning-with-files]. Triggered by harnessed CLI `harnessed task-deliver --task <text>`
+  or slash command `/task-deliver` after `harnessed setup`.
+trigger_phrases:
+  - "deliver this subtask"
+  - "task-deliver workflow"
+  - "Stage 3 deliver"
+  - "ralph-loop COMPLETE"
+  - "跑 task-deliver"
+---
+
+# task-deliver workflow (v3)
+
+## Overview
+
+2-phase sub-workflow mapping the user's CLAUDE.md Stage ③.d 子任务交付 discipline
+onto the harnessed runtime, fully `harnessed.workflow.v3` schema (Phase v3.0-3.4 W0
+T3.4.W0.9 — D-09 L0 Discipline Substrate + D-10 ralph-loop SDK wrapper + D-11 Agent
+Teams 升级 5 触发 OR-chain + R20.10 explicit max_iterations_exceeded handler).
+
+| phase | id | upstream | model | capability / args / parallelism / fallback |
+| ----- | -- | -------- | ----- | ------------------------------------------ |
+| 1 | `01-deliver` | ralph-loop | haiku | `{{ capabilities.ralph-loop.cmd }}` + `args: {completion_promise: COMPLETE, max_iterations: ...}` + `parallelism: judgments.parallelism-gate.ralph-loop-wrapper.fires` + `fallback.max_iterations_exceeded.action: emit_warning_and_halt` |
+| 2 | `02-progress-mark` | planning-with-files | haiku | `{{ capabilities.planning-with-files.cmd }}` / `invokes: /plan` / `artifacts_expected: [progress.md]` |
+
+Per-phase config loads from `workflows/task/deliver/workflow.yaml`; engine.runRouting
+spawns each phase as a sub-agent via `@anthropic-ai/claude-agent-sdk` 0.3.142+.
+
+## Phase 01 ralph-loop COMPLETE wrapper (R20.10 + D-10 + ADR 0011)
+
+ralph-loop SDK wrapper 保 completion-promise verbatim string `"COMPLETE"` — sub-task
+被认为完成的判据是子任务输出包含 verbatim "COMPLETE" string (NOT 启发式 / NOT
+LLM-as-judge). Sister capabilities.yaml `ralph-loop` entry impl `bundled-skill` +
+`sdk_ref: src/routing/lib/ralphLoop.ts` (Phase 2.2 v0.2.0 ship)。
+
+### Parallelism — ralph-loop 正交 wrapper
+
+`parallelism: judgments.parallelism-gate.ralph-loop-wrapper.fires` ref — per R20.10
++ D-10, ralph-loop 是 **正交 wrapper** 套在 subagent-default / agent-teams-upgrade /
+main-session-fallback 任 1 mode 外层 (NOT 互斥触发器, 而是 `wraps:` orthogonal field
+in parallelism-gate.yaml L42-45). Runtime engine 评估 wrapping mode 后 spawn 相应
+execution unit + 套 ralph-loop completion check。
+
+### Agent Teams conditional escalation (D-11 + agent-teams.md 5 OR-chain)
+
+5 升级触发 (per capabilities.yaml `agent-teams-create.fires_when` + agent-teams.md):
+1. `teammate_send_message_needed == true` — teammate 间 SendMessage 互通 (NOT fire-and-forget)
+2. `subagent_context_overflow == true` — subagent 撞 context 上限
+3. `shared_task_list == true` — 多 teammate 共享 task list 自协调
+4. `opposing_hypothesis_debate == true` — 对立假设辩论
+5. `fullstack_three_way == true` — 全栈三路协同
+
+任 1 fire → escalate subagent fan-out → Agent Teams Pattern A/B/C。Cleanup mandatory
+per agent-teams.md 防呆清单 (SendMessage shutdown_request + TeamDelete) — engine-level
+wiring, NOT yaml schema scope。
+
+### R20.10 explicit max_iterations_exceeded handler (NOT silent abort)
+
+phase.fallback.max_iterations_exceeded = `{action: emit_warning_and_halt, message,
+exit_code: 1}` — schema-enforced via FallbackMaxIterationsExceeded Type.Literal(
+'emit_warning_and_halt') (workflow.ts L70-77). Sister Phase 2.4 W1.2 fallbackHandlers.ts
+engine.ts wire — ralph-loop 撞 max_iterations 时 explicit emit warning + halt with
+exit_code 1, NOT silent abort / continue。
+
+Brief enforcement W0.9: ✅ ralph-loop completion_promise COMPLETE / ✅ parallelism-gate
+ref / ✅ R20.10 explicit max_iterations_exceeded handler。
+
+## Phase 02 progress-mark planning-with-files (D-15 + Q-AUDIT-5a Option A)
+
+02-progress-mark invokes Claude Code plugin slash cmd `/plan` to mark subtask complete
+in `progress.md` — sister Phase 01-code progress update pattern, last call in Stage
+③ task chain。Plugin path `~/.claude/plugins/cache/planning-with-files/
+planning-with-files/2.34.0/` verified (2026-05-20)。
+
+## CLI invocation
+
+```bash
+# Dry-run preview — arbitrate-only, never spawns SDK.
+harnessed task-deliver --task "<text>" --dry-run --non-interactive
+
+# Apply path — real SDK spawn + 2-phase chain (ralph-loop COMPLETE + progress mark).
+harnessed task-deliver --task "<text>" --apply
+```
+
+## Forward-looking note
+
+The `trigger_phrases:` frontmatter is active after `harnessed setup` copies this
+SKILL.md to `~/.claude/skills/task-deliver/` — Claude Code then loads the slash
+command `/task-deliver` automatically (Gap B fix — sister v1.0.2 mechanism).
+
+## References
+
+- D-09 — L0 Discipline Substrate always-on (6 disciplines)
+- D-10 — ralph-loop 真接 SDK wrapper (NOT mock reference; v0.2.0 ship)
+- D-11 — Agent Teams 升级 5 触发 OR-chain per ~/.claude/rules/agent-teams.md
+- R20.10 — ralph-loop max_iterations_exceeded explicit emit_warning_and_halt
+  (acceptance c "NOT silent abort"); ralph-loop 正交 wrapper wraps 3 mode
+- D-02 — SKILL.md `name:` bare slash cmd (`task-deliver` NOT `task/deliver`) per ADR 0030
+- ADR 0011 — SDK + ralph-loop integration v0.2.0 baseline
+- ~/.claude/CLAUDE.md "Execute 阶段" 节 ralph-loop 至 verbatim COMPLETE +
+  `子任务并行执行机制` Agent Teams 升级路由
+- ~/.claude/rules/agent-teams.md — Pattern A/B/C + 防呆清单 cleanup mandatory
+- `workflows/judgments/parallelism-gate.yaml` triggers.ralph-loop-wrapper +
+  agent-teams-upgrade + subagent-default + main-session-fallback
+- `workflows/capabilities.yaml` — ralph-loop + agent-teams-{create,send-message,shutdown}
+  + planning-with-files entries
+- `workflows/defaults.yaml` — ralph_max_iterations.task-deliver.* values (T3.4.W2.2 followup)
+- `docs/WORKFLOW.md` — 4-stage workflow mermaid + Stage ③ Execute 章节

package/workflows/task/deliver/workflow.yaml

@@ -0,0 +1,77 @@
+# workflows/task/deliver/workflow.yaml — Phase v3.0-3.4 W0 T3.4.W0.9
+# (D-09 + D-10 + D-11 + R20.10 + ADR 0011 ralph-loop 真接 SDK wrapper)
+#
+# Stage ③.d 子任务交付 — ralph-loop COMPLETE wrapper 保 completion-promise verbatim
+# "COMPLETE" (R20.10 + D-10 v0.2.0 ralph-loop SDK pattern)。Parallelism: subagent
+# default fan-out OR Agent Teams 升级 (5 触发 OR-chain per ~/.claude/rules/agent-teams.md
+# + D-11)。02-progress-mark: planning-with-files /plan 更新 progress.md 标记完成。
+#
+# Brief enforcement (W0.9 spec MUST include):
+#   - ralph-loop completion_promise = COMPLETE (R20.10 verbatim acceptance string)
+#   - parallelism-gate ref: judgments.parallelism-gate.ralph-loop-wrapper.fires
+#     (R20.10 ralph-loop 正交 wraps subagent-default / agent-teams-upgrade / main-session-fallback)
+#   - Agent Teams route conditional via on[] clause when 5 escalation triggers fire
+#   - R20.10 explicit max_iterations_exceeded handler:
+#     {action: emit_warning_and_halt, message, exit_code: 1} — NOT silent abort
+#     (sister Phase 2.4 W1.2 fallbackHandlers.ts engine.ts wire pattern)
+#
+# Sister refs:
+#   - .planning/phase-v3.0-3.2/RESEARCH-workflows.md § Area 2 task/deliver sub yaml example
+#   - workflows/judgments/parallelism-gate.yaml triggers.ralph-loop-wrapper (Phase 2.3 W0.2 SHIPPED)
+#   - workflows/capabilities.yaml ralph-loop + agent-teams-{create,send-message,shutdown} +
+#     planning-with-files (Phase 2.3 W0.1 + 3.3 W0.2 SHIPPED)
+#   - ~/.claude/rules/agent-teams.md 5 升级触发 OR-chain (teammate_send_message_needed /
+#     subagent_context_overflow / shared_task_list / opposing_hypothesis_debate / fullstack_three_way)
+#   - src/workflow/schema/workflow.ts WorkflowSchemaV3 FallbackMaxIterationsExceeded
+#     Type.Literal('emit_warning_and_halt') (Phase 3.3 W0.5 SHIPPED)
+#
+# disciplines_applied: 6 default 全开 per D-09 L0 Discipline Substrate always-on
+# tools_available: 5 entry ∈ capabilities.yaml entry set (T3.3.W0.10 Contract C1 catch)
+
+schema_version: harnessed.workflow.v3
+workflow: task-deliver
+description: |
+  Stage ③.d 子任务交付 — ralph-loop COMPLETE wrapper 保 completion-promise verbatim
+  "COMPLETE" (R20.10 + D-10 v0.2.0 ralph-loop SDK pattern)。Parallelism subagent
+  default fan-out OR Agent Teams 升级 (5 触发 OR-chain per ~/.claude/rules/agent-teams.md
+  + D-11)。R20.10 explicit max_iterations_exceeded handler: emit_warning_and_halt
+  exit_code 1 (NOT silent abort, sister Phase 2.4 W1.2 fallbackHandlers.ts engine.ts
+  wire pattern)。Cleanup mandatory when Agent Teams escalation path fires (shutdown_request
+  + TeamDelete per agent-teams.md 防呆清单)。
+
+disciplines_applied: [karpathy, output-style, language, operational, priority, protocols]
+tools_available:
+  - ralph-loop
+  - agent-teams-create
+  - agent-teams-send-message
+  - agent-teams-shutdown
+  - planning-with-files
+
+phases:
+  - id: 01-deliver
+    name: deliver (ralph-loop COMPLETE wrapper + Agent Teams conditional escalation)
+    upstream: ralph-loop
+    capability: '{{ capabilities.ralph-loop.cmd }}'
+    model: haiku
+    args:
+      completion_promise: COMPLETE
+      max_iterations: '{{ defaults.ralph_max_iterations.task-deliver.01-deliver }}'
+    parallelism: judgments.parallelism-gate.ralph-loop-wrapper.fires
+    on:
+      - if: 'subtask.lines >= 20 and subtask.type != "single_command_query"'
+        invoke: '{{ capabilities.ralph-loop.cmd }}'
+      - if: 'subtask.lines < 20 or subtask.type == "single_command_query"'
+        action: skip
+    fallback:
+      max_iterations_exceeded:
+        action: emit_warning_and_halt
+        message: 'ralph-loop max-iterations exceeded for task-deliver. Sub-task likely incomplete — explicit halt per R20.10 (NOT silent abort).'
+        exit_code: 1
+
+  - id: 02-progress-mark
+    name: progress-mark (planning-with-files /plan — mark subtask complete in progress.md)
+    upstream: planning-with-files
+    capability: '{{ capabilities.planning-with-files.cmd }}'
+    invokes: '/plan'
+    model: haiku
+    artifacts_expected: [progress.md]

package/workflows/task/test/SKILL.md

@@ -0,0 +1,93 @@
+---
+name: task-test
+description: |
+  task-test workflow v3 — Stage ③.c 子任务测试 sub-workflow (superpowers TDD
+  red-green-refactor 强制 + diagnose conditional invoke)。Single-phase composition:
+  01-test (capability `superpowers:test-driven-development` + gate
+  judgments.tdd-gate.tdd-strongly-suggested.fires + invokes_tools[{if: test_fail,
+  tool: diagnose}])。Alias /tdd (mattpocock-skills) 可替代 superpowers TDD per D-13。
+  schema_version: harnessed.workflow.v3 with disciplines_applied [6] + tools_available
+  [tdd, diagnose]. Triggered by harnessed CLI `harnessed task-test --task <text>` or
+  slash command `/task-test` after `harnessed setup`.
+trigger_phrases:
+  - "test this subtask"
+  - "task-test workflow"
+  - "Stage 3 test"
+  - "TDD red-green-refactor"
+  - "跑 task-test"
+---
+
+# task-test workflow (v3)
+
+## Overview
+
+Single-phase sub-workflow mapping the user's CLAUDE.md Stage ③.c 子任务测试 + TDD
+强烈建议开启 discipline onto the harnessed runtime, fully `harnessed.workflow.v3`
+schema (Phase v3.0-3.4 W0 T3.4.W0.8 — D-09 L0 Discipline Substrate + D-04 gate ref
++ D-05 conditional `invokes_tools` + D-13 tdd capability alias).
+
+| phase | id | upstream | model | capability / invokes_tools | gate |
+| ----- | -- | -------- | ----- | -------------------------- | ---- |
+| 1 | `01-test` | superpowers | sonnet | `{{ capabilities.tdd.cmd }}` + `invokes_tools: [{if: test_fail == true, tool: diagnose}]` | `judgments.tdd-gate.tdd-strongly-suggested.fires` |
+
+Per-phase config loads from `workflows/task/test/workflow.yaml`; engine.runRouting
+spawns each phase as a sub-agent via `@anthropic-ai/claude-agent-sdk` 0.3.142+.
+
+## TDD gate (D-04 + judgments/tdd-gate.yaml 6 fires_when + 3 skips_when)
+
+Gate `judgments.tdd-gate.tdd-strongly-suggested.fires` 机器化 CLAUDE.md
+「Execute 阶段」 TDD 强烈建议开启 节 6 OR-chain:
+- `subtask.is_core_business_logic == true`
+- `subtask.is_algorithm == true`
+- `subtask.is_data_processing == true`
+- `subtask.regression_risk == 'high'`
+- `subtask.reliability_required == true`
+
+Skips when (per tdd-gate.yaml skips_when):
+- `subtask.type in ['crud', 'ui_polish', 'docs_only']`
+
+Gate 4-level ref pre-resolved by `judgmentResolver` (T2.3.W0.4 SHIPPED) BEFORE
+expr-eval evaluation — runtime engine 跳过 phase 当 gate 不 fire 时。
+
+## D-13 tdd capability alias
+
+Capability `tdd` (per capabilities.yaml L346-359) 主 impl `superpowers:test-driven-
+development`, alias `[{impl: mattpocock-skills, cmd: /tdd}]` — 两者可替代 per D-13
+LOCKED 决策。`{{ capabilities.tdd.cmd }}` 默认 resolve 至 superpowers
+SDK, 用户 explicit signal 可切换 mattpocock /tdd alias path。
+
+## Conditional diagnose invoke (D-05 invokes_tools)
+
+Phase 01-test 条件性 fire `diagnose` (capabilities.yaml L55-64 mattpocock-skills
+/diagnose) when `test_fail == true` — sister CLAUDE.md "系统化排错" pattern;
+test fail 时进入 diagnose loop (reproduce → minimise → hypothesise → instrument →
+fix → regression-test), 测试通过则 skip diagnose entirely。
+
+## CLI invocation
+
+```bash
+# Dry-run preview — arbitrate-only, never spawns SDK.
+harnessed task-test --task "<text>" --dry-run --non-interactive
+
+# Apply path — real SDK spawn + 1-phase (TDD gate evaluation conditional).
+harnessed task-test --task "<text>" --apply
+```
+
+## Forward-looking note
+
+The `trigger_phrases:` frontmatter is active after `harnessed setup` copies this
+SKILL.md to `~/.claude/skills/task-test/` — Claude Code then loads the slash
+command `/task-test` automatically (Gap B fix — sister v1.0.2 mechanism).
+
+## References
+
+- D-09 — L0 Discipline Substrate always-on
+- D-04 — `gate` 4-level ref pre-resolved by `judgmentResolver`
+- D-05 — phase-level `invokes_tools` conditional tool fire
+- D-13 — tdd capability 2 impl 候选 alias (superpowers 主 + mattpocock /tdd 备)
+- D-02 — SKILL.md `name:` bare slash cmd (`task-test` NOT `task/test`) per ADR 0030
+- ~/.claude/CLAUDE.md "Execute 阶段" TDD 强烈建议开启 6 触发 + 3 跳过
+- `workflows/judgments/tdd-gate.yaml` triggers.tdd-strongly-suggested
+- `workflows/capabilities.yaml` — tdd (superpowers + mattpocock alias) + diagnose entries
+- `workflows/defaults.yaml` — ralph_max_iterations.task-test.* values (T3.4.W2.2 followup)
+- `docs/WORKFLOW.md` — 4-stage workflow mermaid + Stage ③ Execute 章节

package/workflows/task/test/workflow.yaml

@@ -0,0 +1,44 @@
+# workflows/task/test/workflow.yaml — Phase v3.0-3.4 W0 T3.4.W0.8 (D-09 + D-04 + D-05 + D-13)
+#
+# Stage ③.c 子任务测试 — superpowers:test-driven-development red-green-refactor 强制
+# (核心业务逻辑 / 算法 / 数据处理 / 回归 risk high / reliability required 触发 tdd-gate),
+# 其他可选 mattpocock /tdd alias (per D-13 capabilities.yaml `tdd` entry aliases)。
+# Conditional fire: diagnose 当 test_fail == true 时 invoke (D-05 invokes_tools)。
+#
+# Sister refs:
+#   - .planning/phase-v3.0-3.2/RESEARCH-workflows.md § Area 2 task/test sub yaml example
+#   - workflows/judgments/tdd-gate.yaml triggers.tdd-strongly-suggested (Phase 2.3 W0.2 SHIPPED)
+#   - workflows/capabilities.yaml tdd + diagnose (Phase 2.3 W0.1 SHIPPED + D-13 alias)
+#   - ~/.claude/CLAUDE.md "Execute 阶段" 节 TDD 强烈建议开启 6 触发 + 3 跳过
+#   - src/workflow/schema/workflow.ts WorkflowSchemaV3 (Phase 3.3 W0.5 SHIPPED)
+#
+# Brief enforcement:
+#   - W0.8 必带 tdd-gate ref: phase.gate: judgments.tdd-gate.tdd-strongly-suggested.fires
+#   - W0.8 必带 diagnose conditional invoke: invokes_tools[{if: 'test_fail == true', tool: diagnose}]
+#
+# disciplines_applied: 6 default 全开 per D-09 L0 Discipline Substrate always-on
+# tools_available: 2 entry ∈ capabilities.yaml entry set (T3.3.W0.10 Contract C1 catch)
+
+schema_version: harnessed.workflow.v3
+workflow: task-test
+description: |
+  Stage ③.c 子任务测试 — superpowers TDD red-green-refactor 强制 (核心业务 / 算法 /
+  数据处理 / 回归 risk high / reliability required, 6 fires_when OR-chain per
+  ~/.claude/CLAUDE.md), 其他可选 mattpocock /tdd alias。Conditional diagnose
+  invoke when test_fail (D-05 invokes_tools phase-level fire). Gate:
+  judgments.tdd-gate.tdd-strongly-suggested.fires (4-level ref, pre-resolved per D-04)。
+
+disciplines_applied: [karpathy, output-style, language, operational, priority, protocols]
+tools_available: [tdd, diagnose]
+
+phases:
+  - id: 01-test
+    name: test (superpowers TDD red-green-refactor + diagnose conditional invoke)
+    upstream: superpowers
+    capability: '{{ capabilities.tdd.cmd }}'
+    model: sonnet
+    gate: judgments.tdd-gate.tdd-strongly-suggested.fires
+    invokes_tools:
+      - if: 'test_fail == true'
+        tool: diagnose
+    max_iterations: '{{ defaults.ralph_max_iterations.task-test.01-test }}'

package/workflows/verify/auto/SKILL.md

@@ -0,0 +1,77 @@
+---
+name: verify
+description: |
+  Stage ④ Verify master orchestrator — 7 sub conditional per ~/.claude/CLAUDE.md "Verify 阶段"
+  verbatim: progress 必跑 → code-review 并行 → paranoid 关键模块强制 → qa/security/design 可选
+  并行 conditional → simplify 末尾 → multispec 关键发布 Pattern C 4-specialist Agent Team。
+  schema_version: harnessed.workflow.v3 with delegates_to (7 sub: progress serial order 1 +
+  5 parallel conditional + simplify serial order 99) + disciplines_applied (6 default) +
+  tools_available (10 entry)。Triggered by harnessed CLI `harnessed verify --phase <num>` or
+  slash command `/verify` (bare per ADR 0030 namespace policy D-02 LOCK) after `harnessed setup`.
+trigger_phrases:
+  - "verify"
+  - "验证阶段"
+  - "stage 4 verify"
+  - "verify work"
+  - "代码审查 + 简化"
+---
+
+# verify master orchestrator (v3)
+
+## Overview
+
+4-stage cadence Stage ④ master orchestrator delegating to 7 sub-workflows
+(per ~/.claude/CLAUDE.md "Verify 阶段" 节 verbatim — 9-phase composition compressed
+into 7 sub delegation via stage-routing.yaml):
+
+| order/mode | sub | gate ref | mode | when fires |
+| ---------- | --- | -------- | ---- | ---------- |
+| 1 (serial) | `progress` | (unconditional — verify 起点) | serial | always when stage=='verify' |
+| parallel | `code-review` | (unconditional — multi-agent fan-out) | parallel | always |
+| parallel | `paranoid` | `judgments.stage-routing.verify-paranoid-critical.fires` | parallel | phase.is_critical_module == true |
+| parallel | `qa` | `judgments.stage-routing.verify-qa-ui.fires` | parallel | phase.has_ui_changes == true |
+| parallel | `security` | `judgments.stage-routing.verify-security-secrets.fires` | parallel | phase.has_auth_or_secrets == true |
+| parallel | `design` | `judgments.stage-routing.verify-design-changes.fires` | parallel | phase.has_design_changes == true |
+| parallel | `multispec` | `judgments.stage-routing.verify-multispec-critical-release.fires` | parallel | is_critical_release == true (Pattern C 4-specialist Agent Team) |
+| 99 (serial) | `simplify` | (unconditional — 末尾 tail) | serial | always — code-simplifier 末尾移除重复 / 多余逻辑 |
+
+Engine runtime per T3.5.W0.1 `runMasterOrchestrator`:
+- **serial chain**: progress (order 1) 起点 → ... → simplify (order 99) 末尾收尾
+- **parallel fan-out**: 5 conditional sub (code-review + paranoid + qa + security + design + multispec)
+  spawn 并发, 按 gate-eval 结果 fire-or-skip
+- K9 invariant enforced: every serial mode delegate carries explicit `order`
+
+## Verify cadence (sister CLAUDE.md "Verify 阶段" verbatim)
+
+1. 子任务完成后立即 `/gsd-verify-work` + `/gsd-progress` (sub progress 起点必跑串行)
+2. 项目 / 大功能整体完成后:
+   - 先 `code-review` 多 Agent 并行 (sub code-review)
+   - **关键模块强制** `/review` Paranoid Staff Engineer (sub paranoid, gate is_critical_module)
+   - 可选 `/qa` (sub qa, gate has_ui_changes) / `/cso` (sub security, gate has_auth_or_secrets) / `/design-review` (sub design, gate has_design_changes)
+   - **关键发布 / 大重构 PR** 升级 4-specialist Agent Team Pattern C (sub multispec, gate critical-release-upgrade)
+   - 再 `code-simplifier` 末尾 (sub simplify, serial order 99)
+
+## Capability refs
+
+Sister `workflows/capabilities.yaml`:
+- `gsd-verify-work` + `gsd-progress` — Bucket 2 (progress sub upstream)
+- `code-review` + `code-simplifier` — Bucket 1 mattpocock (code-review + simplify subs)
+- `gstack-review` + `gstack-qa` + `gstack-cso` + `gstack-design-review` — Bucket 3 治理关卡 (paranoid/qa/security/design subs)
+- `agent-teams-create` — Bucket 5 agent-platform (multispec Pattern C 4-specialist team)
+- `planning-with-files` — Bucket 4 核心 (progress.md sink throughout)
+
+## Invocation
+
+- CLI: `harnessed verify --phase <num>`
+- Slash command: `/verify` (bare per ADR 0030 namespace policy D-02 LOCK after `harnessed setup`)
+
+## References
+
+- D-01 master orchestrator delegation pattern
+- D-02 bare slash cmd convention (ADR 0030 namespace policy LOCK)
+- D-12 gstack 治理关卡 ref (paranoid / qa / security / design subs)
+- ~/.claude/CLAUDE.md "Verify 阶段" 节 verbatim
+- ~/.claude/rules/agent-teams.md Pattern C 多维度审查 (multispec sub 4-specialist 互相质询)
+- workflows/judgments/stage-routing.yaml — verify-* 6 triggers (7 sub delegation)
+- workflows/verify/{progress,code-review,paranoid,qa,security,design,simplify,multispec}/workflow.yaml
+  — 8 sub-workflow Phase 3.4 SHIPPED

package/workflows/verify/auto/workflow.yaml

@@ -0,0 +1,74 @@
+# workflows/verify/auto/workflow.yaml — Phase v3.0-3.5 W1 T3.5.W1.4
+#
+# Stage ④ Verify master orchestrator — 7 sub conditional per ~/.claude/CLAUDE.md "Verify 阶段"
+# verbatim + RESEARCH-workflows § Area 2 verify master:
+#   progress 必跑 (serial order 1) →
+#   code-review 并行 (parallel) +
+#   paranoid 关键模块 (parallel + gate is-critical-module) +
+#   qa+security+design 可选 conditional (parallel) +
+#   simplify 末尾 (serial order 99) +
+#   multispec 关键发布 Pattern C (parallel + gate critical-release-upgrade)
+#
+# Sister refs:
+#   - ~/.claude/CLAUDE.md "Verify 阶段 (GSD + gstack)" 节 verbatim
+#   - workflows/judgments/stage-routing.yaml — verify-* triggers (7 sub delegation)
+#   - workflows/judgments/parallelism-gate.yaml — agent-teams-upgrade trigger (multispec gate)
+#   - workflows/verify/{progress,code-review,paranoid,qa,security,design,simplify,multispec}/workflow.yaml
+#     — 8 sub-workflow Phase 3.4 SHIPPED
+#   - workflows/capabilities.yaml — gsd-verify-work + gsd-progress + code-review + gstack-review +
+#     gstack-qa + gstack-cso + gstack-design-review + code-simplifier + agent-teams-create +
+#     planning-with-files
+#   - .planning/phase-v3.0-3.2/RESEARCH-workflows.md § Area 2 Master Orchestrator yaml Examples
+#   - src/workflow/schema/workflow.ts WorkflowSchemaV3 + DelegationClause (Phase 3.3 W0.5 SHIPPED)
+#
+# Master invariant (per WorkflowSchemaV3 runtime + check-workflow-schema.mjs K9):
+#   - phases 缺省 (master only has delegates_to[])
+#   - 7 delegates_to: serial 2 (progress order 1 + simplify order 99) + parallel 5
+#   - serial mode 必带 explicit order (K9 invariant T3.3.W0.10)
+#   - parallel mode 无 order field 需求
+
+schema_version: harnessed.workflow.v3
+workflow: verify
+description: |
+  Stage ④ Verify master orchestrator — 7 sub conditional per ~/.claude/CLAUDE.md "Verify 阶段"
+  verbatim: progress 必跑串行 → code-review 并行 → paranoid 关键模块强制 → qa/security/design
+  可选 conditional 并行 → simplify 末尾串行 → multispec 关键发布 Pattern C 4-specialist Agent Team。
+  serial mode 必带 explicit order (K9 invariant); progress order 1 起点, simplify order 99 末尾。
+
+disciplines_applied: [karpathy, output-style, language, operational, priority, protocols]
+tools_available:
+  - gsd-verify-work
+  - gsd-progress
+  - code-review
+  - gstack-review
+  - gstack-qa
+  - gstack-cso
+  - gstack-design-review
+  - code-simplifier
+  - agent-teams-create
+  - planning-with-files
+
+delegates_to:
+  - sub: progress
+    mode: serial
+    order: 1
+  - sub: code-review
+    mode: parallel
+  - sub: paranoid
+    gate: judgments.stage-routing.verify-paranoid-critical.fires
+    mode: parallel
+  - sub: qa
+    gate: judgments.stage-routing.verify-qa-ui.fires
+    mode: parallel
+  - sub: security
+    gate: judgments.stage-routing.verify-security-secrets.fires
+    mode: parallel
+  - sub: design
+    gate: judgments.stage-routing.verify-design-changes.fires
+    mode: parallel
+  - sub: multispec
+    gate: judgments.stage-routing.verify-multispec-critical-release.fires
+    mode: parallel
+  - sub: simplify
+    mode: serial
+    order: 99

package/workflows/verify/code-review/SKILL.md

@@ -0,0 +1,69 @@
+---
+name: verify-code-review
+description: |
+  Stage ④.b verify sub-workflow — code-review 多 agent 并行 fan-out 高置信度 finding
+  (subagent default per ~/.claude/CLAUDE.md 子任务并行机制 — Task / Agent 工具 spawn
+  多 subagent fan-out, context 隔离, token 敏感)。
+  schema_version: harnessed.workflow.v3 with disciplines_applied (6 default) + tools_available
+  (code-review) + 1 phase (parallelism ref judgments.parallelism-gate.subagent-default.fires)。
+  Triggered by harnessed CLI `harnessed verify-code-review --phase <num>` or slash command
+  `/verify-code-review` after `harnessed setup`.
+trigger_phrases:
+  - "verify code review"
+  - "代码审查"
+  - "multi-agent code review"
+  - "并行 review"
+  - "跑 verify-code-review"
+---
+
+# verify-code-review workflow (v3)
+
+## Overview
+
+1-phase sub-workflow mapping CLAUDE.md "Verify 阶段 — code-review 多 agent 并行" 章节
+onto harnessed runtime (Phase v3.0-3.4 W0.11 — D-04 Stage ④ Verify 7 sub + 子任务并行
+机制 subagent default routing 机器化 + Pattern A sub-workflow ship)。
+
+| phase | id | upstream | model | capability | parallelism |
+| ----- | -- | -------- | ----- | ---------- | ----------- |
+| 1 | `01-code-review` | mattpocock-skills | sonnet | `{{ capabilities.code-review.cmd }}` | `judgments.parallelism-gate.subagent-default.fires` |
+
+Per-phase config loads from `workflows/verify/code-review/workflow.yaml`; engine spawns
+multiple subagent in parallel fan-out (sister `~/.claude/CLAUDE.md` 子任务并行机制 默认 —
+Task / Agent 工具 spawn 多任务并发, context 隔离, summary 折叠回主 context)。
+
+## Capability refs
+
+Sister `workflows/capabilities.yaml` entries:
+- `code-review` — Bucket 1 mattpocock 高频招式 (impl: mattpocock-skills, cmd: /code-review)
+
+## Parallelism gate ref
+
+Sister `workflows/judgments/parallelism-gate.yaml`:
+- `subagent-default.fires` — `subtask.parallel_count <= 3 and subtask.communication_needed == false`
+  (默认 fan-out, focused 任务 research / verify / review 单文件 / 跑测试 / 抓 doc / 探索模块)
+
+## Routing rules
+
+总 fire 当 `phase.stage == 'verify'` 后必跑串行 (verify-progress) 之后并行 fan-out。无 skip
+条件 — code-review 多 agent 是 verify-work 第 3 phase 默认 fan-out (sister CLAUDE.md verbatim)。
+
+## CLI invocation
+
+```bash
+# Dry-run preview — arbitrate-only, never spawns SDK.
+harnessed verify-code-review --phase <num> --dry-run --non-interactive
+
+# Apply path — real SDK spawn + parallel fan-out subagent.
+harnessed verify-code-review --phase <num> --apply
+```
+
+## References
+
+- D-04 Stage ④ Verify 7 sub 分解
+- ~/.claude/CLAUDE.md "Verify 阶段 — code-review 多 agent 并行" verbatim
+- ~/.claude/CLAUDE.md "子任务并行执行机制 — subagent vs Agent Teams 路由" subagent default
+- workflows/capabilities.yaml — code-review
+- workflows/judgments/parallelism-gate.yaml — subagent-default.fires
+- workflows/defaults.yaml — ralph_max_iterations.verify-code-review.* values (W2.2 backfill)
+- workflows/verify-work/workflow.yaml v2 SHIPPED phase 03-code-review-parallel sister verbatim

package/workflows/verify/code-review/workflow.yaml

@@ -0,0 +1,32 @@
+# workflows/verify/code-review/workflow.yaml — Phase v3.0-3.4 W0 T3.4.W0.11
+#
+# Stage ④.b verify sub-workflow — code-review 多 agent 并行 fan-out 高置信度 finding
+# (subagent default per CLAUDE.md 子任务并行机制 — Task / Agent 工具 spawn 多任务 fan-out,
+# context 隔离, token 敏感)。
+#
+# Sister refs:
+#   - ~/.claude/CLAUDE.md "Verify 阶段" — "code-review 多 agent 并行" 章节
+#   - ~/.claude/CLAUDE.md "子任务并行执行机制 — subagent vs Agent Teams 路由" subagent default
+#   - workflows/judgments/parallelism-gate.yaml — subagent-default.fires 默认 fan-out
+#   - workflows/capabilities.yaml — code-review (Bucket 1 mattpocock 高频招式)
+#   - workflows/verify-work/workflow.yaml v2 SHIPPED phase 03-code-review-parallel sister pattern
+#   - .planning/phase-v3.0-3.2/RESEARCH-workflows.md § Area 2 verify/code-review example verbatim
+
+schema_version: harnessed.workflow.v3
+workflow: verify-code-review
+description: |
+  Stage ④.b code-review 多 agent 并行 fan-out 高置信度 finding (subagent default per
+  CLAUDE.md 子任务并行机制)。parallelism ref judgments.parallelism-gate.subagent-default.fires
+  机器化 — Task / Agent 工具 spawn 多 subagent 并行 fan-out, context 隔离, token 敏感场景默认。
+
+disciplines_applied: [karpathy, output-style, language, operational, priority, protocols]
+tools_available: [code-review]
+
+phases:
+  - id: 01-code-review
+    name: code-review (multi-agent 并行 fan-out 高置信度 finding)
+    upstream: mattpocock-skills
+    capability: '{{ capabilities.code-review.cmd }}'
+    model: sonnet
+    parallelism: judgments.parallelism-gate.subagent-default.fires
+    max_iterations: '{{ defaults.ralph_max_iterations.verify-code-review.01-code-review }}'