create-yss-spec 3.3.1 → 3.3.4

package/template/.cursor/skills/triage/SKILL.md

@@ -6,7 +6,9 @@ disable-model-invocation: true
 
 # Triage
 
-Move issues on the project issue tracker through a small state machine of triage roles.
+显式处理项目 tracker 中的 Issue / 外部 PR。普通对话由 `yss-product-lifecycle` 分诊；本技能不作为自动路由入口。
+
+在 YSS 仓库中先回交生命周期核对仓库身份、请求范围和当前状态；问题理解及补充信息使用 `orchestration-contract.yaml.request_triage`（位于生命周期技能的 `references/`）。本技能负责取证和建议，不创建正式生命周期资产、不批准合同、不直接改变 Ticket 状态。状态变更回交生命周期核验；远端评论、关闭或其他外部动作另检查明确授权。模板源不生成产品 Ticket。
 
 If this repo treats external pull requests as a request surface (see the issue-tracker config), triage covers them too: **a PR is an issue with attached code** — same roles, same states, same machine, with a few deltas marked "for a PR" below. Resolve a bare `#42` to an issue or PR per the tracker config.
 
@@ -21,28 +23,22 @@ Every comment or issue posted to the issue tracker during triage **must** start
 - [AGENT-BRIEF.md](AGENT-BRIEF.md) — how to write durable agent briefs
 - [OUT-OF-SCOPE.md](OUT-OF-SCOPE.md) — how the `.out-of-scope/` knowledge base works
 
-## Roles
+## Categories and states
 
-Two **category** roles:
+Two categories:
 
 - `bug` — something is broken
 - `enhancement` — new feature or improvement
 
-Five **state** roles:
+五态语义由 `docs/agents/triage-labels.md` 定义：
 
 - `needs-triage` — maintainer needs to evaluate
 - `needs-info` — waiting on reporter for more information
-- `ready-for-agent` — fully specified, ready for an AFK agent
-- `ready-for-human` — needs human implementation
+- `ready-for-agent` — 仅限生命周期确认合同已批准且当前、适用门禁通过的垂直切片。
+- `ready-for-human` — 资产仍需指定数字人或生物人会签，不代表可实现或可合并。
 - `wontfix` — will not be actioned
 
-For a PR, the same states read against the attached code: `ready-for-agent` means a brief is attached and an agent should take the next step on the diff; `ready-for-human` means it's ready for a human to merge.
-
-Every triaged issue should carry exactly one category role and one state role. If state roles conflict, flag it and ask the maintainer before doing anything else.
-
-These are canonical role names. The actual label strings used in the issue tracker may differ. The mapping should have been provided to you. If not, tell the user to run `/setup-matt-pocock-skills`.
-
-State transitions: an unlabeled issue normally goes to `needs-triage` first; from there it moves to `needs-info`, `ready-for-agent`, `ready-for-human`, or `wontfix`. `needs-info` returns to `needs-triage` once the reporter replies. The maintainer can override at any time — flag transitions that look unusual and ask before proceeding.
+PR 附有代码也不能凭简报直接成为 `ready-for-agent`，或凭 `ready-for-human` 推断可合并。读取已配置 tracker 的真实状态；冲突回交生命周期处理。配置缺失按生命周期 setup readiness 处理，不从 Git remote 推断平台。
 
 ## Invocation
 
@@ -67,27 +63,19 @@ Show counts and a one-line summary per item. Let the maintainer pick.
 
 ## Triage a specific issue or PR
 
-1. **Gather context.** Read the full issue or PR (body, comments, labels, author, dates; for a PR, the diff too). Parse any prior triage notes so you don't re-ask resolved questions. Explore the codebase using the project's domain glossary, respecting ADRs in the area. Run two checks against the codebase: (a) **redundancy** — search for an existing implementation of the requested behavior by domain concept (not just the request's wording), and report where you looked. If found, it's an already-implemented `wontfix` (step 5). (b) **prior rejection** — read `.out-of-scope/*.md` and surface any that resembles this request.
+1. **Gather context.** Read the full issue or PR (body, comments, labels, author, dates; for a PR, the diff too). Parse any prior triage notes so you don't re-ask resolved questions. Explore the codebase using the project's domain glossary, respecting ADRs in the area. Run two checks against the codebase: (a) **redundancy** — search for an existing implementation of the requested behavior by domain concept (not just the request's wording), and report where you looked. If found, verify that it satisfies this user scenario before recommending already-implemented `wontfix` (step 5). (b) **prior rejection** — read `.out-of-scope/*.md` and surface any that resembles this request.
 
-2. **Recommend.** Tell the maintainer your category and state recommendation with reasoning, plus a brief codebase summary relevant to the request — including whether it's already implemented. Wait for direction.
+2. **Recommend.** 给出类别、状态建议及证据依据；在已有授权内继续取证，只有影响下一动作的缺口才追问，不为每一步重复索要确认。
 
 3. **Verify the claim.** Before any grilling, check that the claim holds up. For a bug, reproduce it from the reporter's steps. For a PR, confirm the diff does what it claims — check it out, run the relevant tests or commands. Report what happened: confirmed (with code path), failed, or insufficient detail (a strong `needs-info` signal). A confirmed verification makes a much stronger agent brief.
 
-4. **Grill (if needed).** If the request needs fleshing out, call the Skill tool twice, for "grilling" and "domain-modeling", and grill it into shape a round of questions at a time, sharpening domain terms and updating `CONTEXT.md`/ADRs inline as decisions land.
+4. **Grill (if needed).** If the request needs fleshing out, call the Skill tool twice, for "grilling" and "domain-modeling", and grill it into shape a round of questions at a time, sharpening domain terms; return proposed terminology and ADR changes to the lifecycle orchestrator before formal writes.
 
-5. **Apply the outcome:**
-   - `ready-for-agent` — post an agent brief comment ([AGENT-BRIEF.md](AGENT-BRIEF.md)).
-   - `ready-for-human` — same structure as an agent brief, but note why it can't be delegated (judgment calls, external access, design decisions, manual testing).
-   - `needs-info` — post triage notes (template below).
-   - `wontfix` — close, with the comment depending on *why*:
-     - **Already implemented** — the change already exists in the codebase. Point to where it lives; do **not** write to `.out-of-scope/` (that KB is for *rejected* requests, not built ones).
-     - **Rejected (bug)** — polite explanation, then close.
-     - **Rejected (enhancement)** — write to `.out-of-scope/`, link to it from a comment, then close ([OUT-OF-SCOPE.md](OUT-OF-SCOPE.md)).
-   - `needs-triage` — apply the role. Optional comment if there's partial progress.
+5. **Return the outcome.** 返回类别与状态建议、已验证事实、未决项、证据、逐目标下一动作，以及 [AGENT-BRIEF.md](AGENT-BRIEF.md) 格式的简报草案。`needs-info` 只询问真正缺少的信息；`wontfix` 必须区分已实现与拒绝，已有实现还需确认符合用户场景，不能仅因存在同名功能就关闭。拒绝 enhancement 时可提出 `.out-of-scope/` 更新建议；正式写入与 tracker 动作交由生命周期按范围和授权执行。
 
-## Quick state override
+## State change requests
 
-If the maintainer says "move #42 to ready-for-agent", trust them and apply the role directly. Confirm what you're about to do (role changes, comment, close), then act. Skip grilling. If moving to `ready-for-agent` without a grilling session, ask whether they want to write an agent brief.
+用户明确要求把 #42 改为 `ready-for-agent` 时，把该请求交给生命周期复算就绪条件；简报、口头要求和 PR diff 均不能代替当前批准合同与门禁证据。缺失时说明具体阻塞，不直接改标签。其他状态也按 tracker 权威规则核验后执行，不重新定义五态。
 
 ## Needs-info template
 

package/template/.cursor/skills/yss-antd-design/references/evidence.md

@@ -25,11 +25,11 @@ JSON 输出存为相邻文件，证据清单只引用路径。不要把整份 `l
 
 `antd lint` 只对可解析的 React/TSX 源码有规则命中。HTML 原型会出现空 issues（假绿）或 `skippedFiles` / `parse-error` / `partial: true`。
 
-CLI JSON 只能证明查询事实，不能单独证明原型消费了目标版本或完成了 Vue/YSS 映射。`gate.prototype-verified` 还必须通过 schema v3 档位证据、Design QA、浏览器和无障碍验证。
+CLI JSON 只能证明查询事实，不能单独证明原型消费了目标版本或完成了 Vue/YSS 映射。`check.prototype-verified` 还必须通过 schema v3 档位证据、Design QA、浏览器和无障碍验证。
 
 | 产物 | `lint_applicable` | 做法 |
 |---|---|---|
 | 静态 HTML / H1 | `not-applicable` | 不创建 AntD 段；由 H1 合同验证 HTML |
 | React/TSX 原型源 | `true` | 对变更文件跑 `antd lint --format json`，保存输出；无 issue 才把 `lint_passed` 设为 `true` |
 
-浏览器验证仍由 `yss-prototype-stage` 采集，不在本技能宣布 `gate.prototype-verified` 通过。
+浏览器验证仍由 `yss-prototype-stage` 采集，不在本技能宣布 `check.prototype-verified` 通过。

package/template/.cursor/skills/yss-antdv-next-design/SKILL.md

@@ -43,6 +43,6 @@ node .agents/skills/yss-antdv-next-design/scripts/collect-antdv-next-facts.mjs \
 - 禁止 `npx skills add antdv-next/cli`、`npx skills add antdv-next/skills`，也不得把官方 Skill 复制进任何 Agent root。
 - 禁止把 MCP 对话、`llms-full.txt`、CLI 自报的未解析版本或静态 `design.md` 当作 exact component API 证据。
 - 禁止让上游 `design.md` 覆盖项目 `DESIGN.md`、`docs/design/design.md` 或项目 Token；manifest 必须直接记录根 `DESIGN.md` path 与 digest。
-- 禁止把原型事实包当作生产兼容性、可复用代码、`gate.prototype-verified` 或组件库迁移完成证据。
+- 禁止把原型事实包当作生产兼容性、可复用代码、`check.prototype-verified` 或组件库迁移完成证据。
 
 事实包合同见 `references/fact-pack.md`，阶段及生产边界见 `references/boundaries.md`。

package/template/.cursor/skills/yss-implementation-contract-compiler/SKILL.md

@@ -39,7 +39,7 @@ description: Use when a YSS vertical slice is entering implementation, spans mul
 - Harness 内实现路径必须落在 `apps/backend/<project>/` 或 `apps/frontend/<project>/` 的具体项目目录；`apps/backend/`、`apps/frontend/` 只能作为容器，`app/backend/`、`app/frontend/` 及其子路径一律阻断。外部实现仓库使用其登记的真实项目根路径。`git-submodule` 使用 `implementation_path_policy: git-submodule-harness-apps`，空 gitlink、detached HEAD 或 `--force` 覆盖挂载点不得脚手架；`inspectWorkingTreeScope.writable` 必须为显式布尔值。
 - 当前用户、缓存、审计、Excel、分布式 ID、请求校验、错误映射、加解密或网关韧性命中时，必须按 `compiler-contract.yaml` 的 `impact_to_capabilities` 补齐入口 capability；不能只在 `boundaries.md` 中提及。仅复用已经验证的平台认证 / 授权能力不算 component impact，不自动增加权限专项 skill。
 - 业务行为使用 `behavior-tdd`；只有机械脚手架/生成物可用 `controlled-generation`，并记录例外和验证。
-- 阶段 5 若命中交付面且无已有工程，按 Project Scaffold Contract schema v4 编译 `controlled-generation` 合同 draft。后端先读取 `scaffold-architecture-decisions.yaml`，只接受已达到 `lifecycle-approved` 且 digest 当前的 DDD / Layered MVC 选择；前端绑定标准模板精确 commit、应用标识、路由、OpenAPI 影响、目标路径和 `init_git` 决定。合同经生命周期批准并持久化后才能运行生成器。历史后端 schema v3 只读兼容并可完成已批准生成；前端只接受 v4。生成与验证完成、仓库登记并通过 `gate.implementation-repositories-ready` 后，才编译 Ticket 与业务 Slice Contract。
+- 阶段 5 若命中交付面且无已有工程，按 Project Scaffold Contract schema v4 编译 `controlled-generation` 合同 draft。后端先读取 `scaffold-architecture-decisions.yaml`，只接受已达到 `lifecycle-approved` 且 digest 当前的 DDD / Layered MVC 选择；前端绑定标准模板精确 commit、应用标识、路由、OpenAPI 影响、目标路径和 `init_git` 决定。合同经生命周期批准并持久化后才能运行生成器。历史后端 schema v3 只读兼容并可完成已批准生成；前端只接受 v4。生成与验证完成、仓库登记并通过 `check.implementation-repositories-ready` 后，才编译 Ticket 与业务 Slice Contract。
 - 脚手架合同在编译阶段只能是 `draft` / `ready-for-lifecycle-review` / `blocked`；只有生命周期编排器可以把已持久化脚手架合同标记为 `approved`。脚手架合同只覆盖业务代码前的工程骨架工作单元；生成、基线校验和合同重编译完成后，它不能替代脚手架后的 Slice Implementation Contract。
 - 脚手架输出消费批准的脚手架合同；脚手架后的所有生成后端代码都必须绑定当前批准且版本当前的 Slice Implementation Contract、主 YSS skill、依赖闭包、允许写路径、预期证据和 YSS Skill Execution Result。打印命令、`./mvnw validate` 单项通过或脚手架成功不能替代合同批准；生成范围从机械内容变成业务行为时触发完整重路由。
 - 专项结果中的越界路径、缺失证据、`drift`、`violation` 或 `new_impacts` 必须阻断或重路由。

package/template/.cursor/skills/yss-mvc-data-analysis-project-initializer/scripts/generate_project.mjs

@@ -23,7 +23,7 @@ async function sharedTools(outputDir, contract) {
   }
   const staging = await mkdtemp(path.join(outputDir, ".skillUtils.staging-"));
   try {
-    for (const relative of [".agents/skills", ".codex/skills", ".claude/skills", ".cursor/skills", ".pi/skills", ".qoder/skills", ".trae/skills", "scripts", "docs/agents", "docs/process", "AGENTS.md", "skills-lock.json"]) {
+    for (const relative of [".agents/skills", ".codex/skills", ".cursor/skills", ".pi/skills", "scripts", "docs/agents", "docs/process", "AGENTS.md", "skills-lock.json"]) {
       await mkdir(path.dirname(path.join(staging, relative)), { recursive: true });
       await cp(path.join(root, relative), path.join(staging, relative), { recursive: true });
     }

package/template/.cursor/skills/yss-mvc-scaffold-generator/scripts/lib/runtime.mjs

@@ -11,8 +11,7 @@ export const HARNESS_ROOT = findHarnessRoot();
 export const SHARED_SKILLS_ROOT = path.join(HARNESS_ROOT, ".agents", "skills");
 export const SKILL_UTILS_NAME = "skillUtils";
 export const SKILL_UTILS_DIRECTORIES = [
-  ".agents/skills", ".claude/skills", ".codex/skills", ".cursor/skills",
-  ".pi/skills", ".qoder/skills", ".trae/skills"
+  ".agents/skills", ".codex/skills", ".cursor/skills", ".pi/skills"
 ];
 export const PROJECT_SCRIPT_FILES = [
   "check-agent-environment.mjs", "implementation-path-policy", "repository-mode", "repository-scope-policy",

package/template/.cursor/skills/yss-mvc-scaffold-generator/scripts/lib/storage.mjs

@@ -57,7 +57,7 @@ async function populateSkillUtils(destination) {
   const publicSkills = path.join(HARNESS_ROOT, "yss-public-skills.json");
   if (!await exists(publicSkills)) fail(`技能工具包源文件不存在: ${publicSkills}`);
   await cp(publicSkills, path.join(destination, "yss-public-skills.json"));
-  await writeFile(path.join(destination, "skill-utils.yaml"), `schema_version: 1\nkind: yss-skill-utils\ntool_version: 1.0.0\ncompatibility: skill-utils-v1\nsource: ${path.basename(HARNESS_ROOT)}\ncanonical_root: .agents/skills\nprojection_roots:\n  - .codex/skills\n  - .claude/skills\n  - .cursor/skills\nexcluded_skills:\n  - yss-mvc-scaffold-generator\n`, "utf8");
+  await writeFile(path.join(destination, "skill-utils.yaml"), `schema_version: 1\nkind: yss-skill-utils\ntool_version: 1.0.0\ncompatibility: skill-utils-v1\nsource: ${path.basename(HARNESS_ROOT)}\ncanonical_root: .agents/skills\nprojection_roots:\n  - .codex/skills\n  - .cursor/skills\n  - .pi/skills\nexcluded_skills:\n  - yss-mvc-scaffold-generator\n`, "utf8");
 }
 
 export async function ensureSkillUtils(targetDir, { apply = true } = {}) {

package/template/.cursor/skills/yss-mvc-scaffold-generator/scripts/verify_project.mjs

@@ -15,11 +15,11 @@ try {
   if (manifest.runtime_java !== "8" || manifest.project_version !== "2.0.0-SNAPSHOT" || manifest.persistence_profile !== "yss-mybatis-plus") throw new Error("Java 8/YSS MyBatis-Plus 技术基线不正确");
   const skillUtils = path.resolve(root, manifest.skill_utils_dir);
   await Promise.all(["skill-utils.yaml", "skills-lock.json", ".agents/skills/yss-product-lifecycle/SKILL.md", ".agents/skills/yss-implementation-contract-compiler/SKILL.md"].map((relative) => required(path.join(manifest.skill_utils_dir, relative))));
-  for (const projection of [".agents", ".claude", ".codex", ".cursor", ".pi", ".qoder", ".trae"]) {
+  for (const projection of [".agents", ".codex", ".cursor", ".pi"]) {
     const nestedGenerator = path.join(skillUtils, projection, "skills", "yss-mvc-scaffold-generator");
     if (await stat(nestedGenerator).then(() => true).catch(() => false)) throw new Error(`skillUtils 不应分发创建期生成器: ${projection}/skills/yss-mvc-scaffold-generator`);
   }
-  for (const forbidden of [".agents", ".claude", ".codex", ".cursor", ".pi", ".qoder", ".trae"]) if (await stat(path.join(root, forbidden)).then(() => true).catch(() => false)) throw new Error(`项目不应携带技能投影目录: ${forbidden}`);
+  for (const forbidden of [".agents", ".codex", ".cursor", ".pi"]) if (await stat(path.join(root, forbidden)).then(() => true).catch(() => false)) throw new Error(`项目不应携带技能投影目录: ${forbidden}`);
   if (JSON.stringify(manifest.modules) !== JSON.stringify(modules)) throw new Error("模块集合或顺序不正确");
   const backendRoot = path.join(root, manifest.backend_root || "");
   await Promise.all(["pom.xml", "mvnw", "mvnw.cmd", ...modules.map((item) => `${item}/pom.xml`)].map((relative) => required(path.join(manifest.backend_root || "", relative))));

package/template/.cursor/skills/yss-openapi-draft-review/SKILL.md

@@ -10,6 +10,7 @@ Use this skill after OpenAPI Draft creation and before Engineering Baseline / YS
 ## Required Inputs
 
 - OpenAPI Draft under `docs/.scratch/<feature>/api/<feature>.yaml`，作为唯一权威的单一 OAS 3.1 YAML document；生命周期元数据和 Freeze 决策位于相邻 Markdown 记录。
+- `docs/.scratch/<feature>/api/<feature>-validation.yaml`，且已由 `scripts/verify-openapi-draft-validation-record` 针对当前 Draft SHA-256 验证通过。
 - Calibrated Spec.
 - Interaction spec / prototype review when UI exists.
 - YSS engineering baseline rules and `.agents/skills/yss-dto/references/openapi-wire-profile.yaml`；先运行 `scripts/verify-yss-dto-openapi-profile`。
@@ -17,10 +18,11 @@ Use this skill after OpenAPI Draft creation and before Engineering Baseline / YS
 
 ## Review Flow
 
-1. Consume fresh automated evidence for YAML parsing, single-document OAS 3.1 shape, `$ref`, path parameters, and lint. Generate it once with the Draft or when the Draft changes; do not manually repeat unchanged structural checks.
-2. Build a P0 traceability matrix from Spec functional requirements and interaction actions to OpenAPI paths, schemas, errors, and contract tests. Each UI action must map to a stable `operationId` and `x-yss-action-key` or an equivalent traceability entry.
+1. Recompute the Draft SHA-256 and verify `<feature>-validation.yaml`. The record must bind a pnpm-lockfile-pinned Redocly version, the exact lint command, exit code, execution time and readable evidence; its structural checks and lint must all pass. Custom parser evidence may supplement this record but cannot replace locked Redocly lint.
+2. Build a P0 traceability matrix from Spec functional requirements and interaction actions to OpenAPI paths, schemas, errors, and contract tests. Each UI action must map to a stable `operationId` and `x-yss-action-key` or an equivalent traceability entry. For P0 write/configuration models and read models that affect a key interaction, trace the exact property path, type/nesting, operation-specific requiredness, nullable/default/enum/format rules, source locator and error/test seam.
 3. Check page action coverage: every action has `actionKey`, endpoint or explicit non-goal, state transition, idempotency/concurrency rule, and error codes. When the Spec explicitly changes authentication or authorization behavior, trace that behavior through the same matrix.
 4. Check object lifecycle coverage: manage/maintain/configure/create/update/archive/retry/cancel/publish/export/create-draft semantics have endpoints or explicit scope downgrades.
+   When create/update or other lifecycle operations share one schema, verify that omission and requiredness semantics match for every operation; otherwise require split schemas or an explicit conditional contract. For credentials, distinguish create, update, masked readback, omit-to-preserve and explicit-clear behavior instead of imposing a universal required rule.
 5. Check YSS API baseline: REST shape, `x-yss-response-wrapper` values `SingleResult|MultiResult|PageResult`, `YssResultMeta` plus `allOf` concrete schemas, and stable DTO/schema names. The Java generic notation is a review shorthand only; it must not appear as an OpenAPI type or `$ref`.
 6. Check DTO wire shape against the profile: `success` is boolean, `dataType` is `string|null`, `code` is only `string|integer|null`, list/page `data` is an array, and single `data` points to the endpoint schema with explicit nullability. Check response/request direction separately.
 7. Check pagination input: only `pageIndex/pageSize/orderBy/orderDirection/groupBy` are client fields; `orderDirection` is `ASC|DESC`; `orderBy` / `groupBy` use endpoint whitelists; `offset`, `needTotalCount`, and `tempTotalCount` are negative assertions. `totalPages` is mapper-dependent and cannot enter a shared schema without fresh wire evidence.
@@ -30,6 +32,7 @@ Use this skill after OpenAPI Draft creation and before Engineering Baseline / YS
 ## Automation Boundary
 
 - Automated checks own YAML single-document syntax, `$ref` resolution, path-parameter consistency, OpenAPI lint, and stable machine-checkable style rules.
+- `scripts/verify-openapi-draft-validation-record` owns validation-record shape, current-byte SHA binding, locked toolchain evidence and deterministic structural rechecks. It does not decide whether a business property correctly implements the Spec.
 - `scripts/verify-yss-dto-openapi-profile` owns the reusable DTO mapping invariants; the review still checks each endpoint's concrete schema, wrapper extension and evidence.
 - Human or independent semantic review owns P0 traceability, page-action coverage, error behavior, concurrency/idempotency, scope downgrades, contract-test seams, and any explicitly specified authentication or authorization behavior.
 - Fresh passing automation evidence may be referenced by the semantic review; copying the same findings into a second checklist is unnecessary.
@@ -40,6 +43,9 @@ Use this skill after OpenAPI Draft creation and before Engineering Baseline / YS
 Block if any of these are true:
 
 - A P0 requirement has no endpoint, schema, error contract, or explicit non-goal.
+- The validation record is missing, stale, blocked or unverifiable; locked Redocly lint is absent or non-zero. Semantic pre-review may continue, but the overall Review Result remains `Blocked`.
+- A P0 write/configuration model or key-interaction read model lacks property-level source, nested shape, operation-specific requiredness, constraints, error or test traceability.
+- Multiple lifecycle operations reuse one schema even though requiredness or omission semantics differ and no explicit conditional contract resolves the difference.
 - A UI action has no endpoint/non-goal mapping.
 - A configurable rule or gate lacks a source, owner, fixed/default decision, or API representation.
 - Pagination does not align with the `PageResult` profile or documented exception, or exposes `offset` / `needTotalCount` / `tempTotalCount` as client input.
@@ -54,6 +60,12 @@ Block if any of these are true:
 ### Review Result
 <Approved / Blocked>
 
+### Structural Validation
+<Passed / Blocked; validation record, Draft SHA-256, locked Redocly version, command and evidence>
+
+### Semantic Review
+<Passed / Blocked; P0 field traceability and remaining semantic findings>
+
 ### Blocking Findings
 - <file:line grounded finding>
 
@@ -62,6 +74,7 @@ Block if any of these are true:
 
 ### Contract Coverage
 - <P0 requirement -> endpoint/schema/error/test mapping summary>
+- <P0 field source -> operationId/schema/property path/shape/requiredness/constraints/error/test mapping summary>
 
 ### YSS Baseline
 - <`x-yss-response-wrapper` map and `YssResultMeta` / `allOf` conformance>
@@ -79,4 +92,6 @@ Block if any of these are true:
 - <return to OpenAPI Draft / enter Engineering Baseline / architecture design / request YAML-to-JSON export after Freeze>
 ```
 
+`Review Result` 只有在 `Structural Validation` 与 `Semantic Review` 均为 `Passed` 时才能是 `Approved`；不使用 `Conditionally Approved` 或 `Semantic Approved` 作为顶层结论。
+
 Prefer `docs/api/templates/openapi-draft-review-checklist.md` when a tabular checklist is useful.

package/template/.cursor/skills/yss-openapi-governance/SKILL.md

@@ -23,6 +23,7 @@ YSS DTO 的可复用 HTTP/JSON 映射由 `.agents/skills/yss-dto/references/open
 - 保证 YAML 是单一 YAML document、根节点为 `openapi: 3.1.0`，且不把 `pipeline`、`stage`、`status`、`owner` 等生命周期元数据写入 OpenAPI 根节点。
 - 按 `yss-dto` wire profile 校验 `com.yss.cloud.dto.result` canonical 包、`YssResultMeta` 公共字段、具体 wrapper schema、请求 / 响应方向和分页负向字段。
 - 运行受项目 lockfile 约束的 lint / bundle，检查 `$ref`、operationId、响应包装、错误、分页、幂等和契约测试 seam；只有 Spec 明确改变认证或授权行为时才检查对应契约。
+- 在 `docs/.scratch/<feature>/api/<feature>-validation.yaml` 持久化结构校验记录，并用 `scripts/verify-openapi-draft-validation-record` 复核当前 YAML SHA-256、锁定的 Redocly 版本、实际命令、退出码、执行时间和证据引用。
 - 在 OpenAPI Freeze 后，用锁定的 Redocly CLI 将 YAML bundle 为 JSON，并记录可重现证据。
 - 维护治理记录、Freeze 记录和 JSON 派生记录。
 - Spec Delta 影响存在时，在 `docs/.scratch/<feature>/spec-delta/` 记录与冻结 YAML 的关系；没有影响时明确记录 `not-applicable`。
@@ -60,14 +61,18 @@ pnpm exec redocly bundle \
    - 所有操作使用稳定、可生成客户端的 `operationId`；页面动作可通过 `x-yss-action-key` 或同路径的追踪矩阵关联。
 
 2. **运行结构与治理校验**
-   - 先执行项目锁定的 `pnpm exec redocly lint` 或等价 CI 脚本。
+   - 先执行项目 lockfile 锁定版本的 `pnpm exec redocly lint`；CI 可以包装调用，但 validation record 必须记录这条实际命令与退出码。
    - 检查 YAML 可解析、`$ref` 可解析、路径参数完整、operationId 唯一、examples 合法、schema 命名稳定。
+   - 从 `docs/api/templates/openapi-draft-validation-record-template.yaml` 创建 `<feature>-validation.yaml`，记录 YAML SHA-256 与 lint 工具链；运行 `scripts/verify-openapi-draft-validation-record --root <project-root> <record>`。自定义解析脚本可以补充诊断，但不能代替锁定的 Redocly lint evidence。
+   - 对每个 P0 写模型、配置模型及影响关键交互的读模型建立字段级追踪：`Spec/交互来源 → operationId → request/response schema → property path → 类型与嵌套形状 → create/update requiredness → nullable/default/enum/format → error/test seam`。数组元素使用 `items[].property` 一类稳定 property path。
+   - create/update 或其他不同生命周期操作复用同一写 schema 时，逐操作证明 requiredness 与省略语义一致；若不一致，拆分 schema 或明确条件契约。凭据类字段还要说明创建、更新、掩码回显、省略保留与显式清空语义，不能仅凭字段名统一设为必填。
    - 先运行 `scripts/verify-yss-dto-openapi-profile`，并记录 profile 版本；检查 `/api/v1/` 版本策略（或记录例外）、`x-yss-response-wrapper`、`YssResultMeta` + `allOf` 具体 schema、统一错误结构、分页、幂等 / 乐观锁和契约测试 seam。
    - 每个响应都必须落成具体 endpoint schema：`SingleResult` 的 `data` 是具体对象或显式 nullable schema，`MultiResult` / `PageResult` 的 `data` 是数组；Java 的 `SingleResult<T>` / `PageResult<T>` 只能作为语义说明，不能直接写成 OAS type 或 `$ref`。
    - `code` 按 profile 只允许 `string | integer | null`，`dataType` 按 profile 为 `string | null`；`offset`、`needTotalCount`、`tempTotalCount` 不得进入客户端分页输入；`totalPages` 只有目标 HTTP mapper / fixture 证明后才能进入契约。Spec 明确改变认证或授权行为时，把对应 `401` / `403`、资源过滤和错误语义作为普通 API 行为检查。
 
 3. **独立 Draft Review 与 Freeze**
-   - 将 fresh lint 证据交给 `yss-openapi-draft-review`；阻断项未关闭前，YAML 仍是 review-only Draft，不得生成生产客户端。
+   - 将通过 verifier 且 SHA-256 与当前 YAML 一致的 validation record 交给 `yss-openapi-draft-review`。缺锁定 lint 时可以先做语义预审，但独立 Review 总结果必须为 `Blocked`；不能用“Freeze 前补 lint”支持 `Approved`。
+   - YAML、validation record 或 lint ruleset 变化后，旧结构证据与旧 Review 立即失效，必须重新校验和审查。阻断项未关闭前，YAML 仍是 review-only Draft，不得生成生产客户端。
    - Freeze 记录必须引用 YAML 路径、Git ref（如适用）和 YAML SHA-256。冻结后 API 行为变更必须先回到 YAML Draft 与审查。
 
 4. **从冻结 YAML 派生 JSON**
@@ -88,7 +93,9 @@ pnpm exec redocly bundle \
 
 - YAML 不是单一 OAS 3.1 document，或其根节点混入生命周期元数据。
 - YAML / `$ref` / lint 不通过，operationId 不稳定或不唯一，或路径参数、schema、examples 无法解析。
+- `<feature>-validation.yaml` 缺失、verifier 失败、记录的 YAML SHA-256 与当前字节不一致，或 Redocly 未由 pnpm lockfile 锁定。
 - P0 操作缺请求、响应、错误、并发 / 幂等规则或可验证 seam；Spec 明确的认证或授权行为没有契约表示。
+- P0 写模型、配置模型或关键交互读模型没有追到 property path、嵌套形状、requiredness、约束、来源和 test seam；跨操作复用 schema 却没有证明省略与必填语义一致。
 - `$ref` 超出允许范围，或转换器版本、lockfile、命令、输入 YAML 无法识别。
 - `scripts/verify-yss-dto-openapi-profile` 失败、profile 版本未记录，或 Draft 未按 profile 建模 `x-yss-response-wrapper`、`YssResultMeta` / `allOf` 和具体 `data` schema。
 - 新契约引入 `com.yss.cloud.dto.response`、把 Java 泛型文字当成 OAS schema、把全局 `code` 放宽为任意 object，或把 `offset` / `needTotalCount` / `tempTotalCount` 作为客户端字段。
@@ -110,7 +117,9 @@ pnpm exec redocly bundle \
 
 ### Validation
 - Lint command and result: <locked pnpm command / result>
+- Validation record: <docs/.scratch/<feature>/api/<feature>-validation.yaml / verifier result / matching YAML SHA-256>
 - YSS DTO wire profile: <`.agents/skills/yss-dto/references/openapi-wire-profile.yaml`, schema_version, verifier result>
+- P0 field traceability: <source / operationId / schema / property path / shape / requiredness / constraints / error-test summary>
 - Wrapper conformance: <`x-yss-response-wrapper`, `YssResultMeta`, `allOf`, concrete data schema, direction and forbidden-field result>
 - `$ref` policy / approved exceptions: <details>
 - Blocking findings: <file:line grounded finding>

package/template/.cursor/skills/yss-product-lifecycle/SKILL.md

@@ -5,15 +5,15 @@ description: 编排 YSS 产品或模块从机会调研到 Spec、原型、技术
 
 # YSS Product Lifecycle
 
-生命周期的单一主控入口。它识别仓库身份和最近可信阶段，计算可执行 frontier，选择并验收下一个工作单元；业务设计、实现和独立审查交给相应专项 skill。主控不替代专项 skill，也不复制它们的执行手册。
+生命周期单一主控：识别请求、仓库身份和最近可信阶段，计算可执行 frontier，派发并验收工作单元；业务设计、实现和独立审查交给专项 skill。
 
 ## 事实源与按需加载
 
-先读取项目根 `yss-project.yaml`、`CONTEXT.md`、相关 ADR、父 Ticket/checkpoint 和当前资产。随后只加载当前决策需要的合同子树：
+先读 `yss-project.yaml`、`CONTEXT.md`、相关 ADR、父 Ticket/checkpoint 和当前资产，按需加载合同子树：
 
 | 事实 | 权威来源 |
 |---|---|
-| 阶段、门禁、产物、工作单元、证据和稳定 ID | `docs/process/lifecycle-registry.yaml` |
+| 阶段、聚合门禁、内部检查、产物、工作单元、证据和稳定 ID | `docs/process/lifecycle-registry.yaml` |
 | 执行模式、流转、readiness、暂停、结果和授权语义 | `references/orchestration-contract.yaml` |
 | Skill 身份、capability、依赖和路由 | `docs/agents/yss-skill-registry.yaml` |
 | 数字人角色、运行时和会签策略 | `docs/agents/digital-human-roles.yaml` |
@@ -33,11 +33,13 @@ scripts/query-lifecycle-context --mode route --stage stage.plan --work-unit work
 
 1. 严格解析 `yss-project.yaml`，不得按目录、Git remote 或占位符猜仓库身份。
 2. 判定影响面和最近可信阶段；阶段完成必须同时满足内容、审查结论、上游新鲜度和可读证据，文件存在不代表通过。
-3. 模式为 `route`、`orchestrate`、`resume`、`audit`；未明确时使用只读 `route`。具体写入边界从编排合同的 `modes` 子树读取。
+3. 按合同 `request_triage` 选择 `route`、`orchestrate`、`resume`、`audit`；明确行动请求无需模式关键字，意图不明时只读 `route`。`modes`、门禁及授权边界仍适用。
 4. `project-instance` 按生命周期注册表推进；`template-source` 只走模板维护流程。
 
 Plan 入口读取 `docs/plan/README.md` 和 `docs/process/plan-migration.md`，按注册表的退出条件核查战略输入。关键未决项阻断进入 Spec；非关键细节须有责任人、解决时点和下游接收方。只使用 Plan 标识；旧阶段不提供兼容解析，历史批准不自动成为 Plan 批准。
 
+理解、澄清与纠错见 [请求分诊协议](references/request-triage.md)；用 `scripts/query-lifecycle-context --include request_triage` 查询合同。先查上下文和证据，只问影响下一动作的缺口，不按问题长度追问。
+
 ## 不可越过的边界
 
 ### 仓库身份
@@ -48,6 +50,8 @@ Plan 入口读取 `docs/plan/README.md` 和 `docs/process/plan-migration.md`，
 
 不得越过命中的阶段、门禁、实现仓库准备或 Ticket 正式化。实现只接收绑定垂直切片、已批准且持久化、版本当前并通过完整 `ready-for-agent` 计算的合同；父 Ticket、`ready-for-human` 切片、`stale`、`drift`、`new_impacts`、`violation` 或缺失证据均阻断。实现仓库、脚手架、UI 还原、review input 和发布条件从对应合同子树查询，不在入口重复定义。
 
+聚合批准见 `gate_consolidation`。
+
 ### 用户决定
 
 命中 `docs/agents/digital-human-roles.yaml.user_decision_policy` 的关键决定，必须先展示当前资产、版本、变化、风险、推荐方案、批准范围和后续动作，再取得提问者或其明确指定生物人负责人的原始回复。数字人同意、超时、默认项或无反对意见不能代答；恢复前按 `references/user-decisions.md` 验证当前资产与原始回复绑定。
@@ -62,7 +66,7 @@ Plan 入口读取 `docs/plan/README.md` 和 `docs/process/plan-migration.md`，
 2. 评估影响面、上游新鲜度、门禁和阻塞，选择第一个未阻塞工作单元；不把 `not-applicable` 当作豁免。
 3. 原生工作单元由主控持有正式资产；专项工作通过结构化任务包派发，只调用编排合同允许的 model-invoked skill。
 4. Matt 的 `grill-with-docs`、`to-spec`、`to-tickets`、`implement` 等仅为显式 user-invoked 兼容入口。主控不得自动调用它们或代替其创建正式资产；Matt 仅导航，不得写生命周期资产或改变门禁/Ticket 状态，任何写入前回交本编排器。
-5. 验收 `Workflow Execution Result`：工作单元、合同、允许写路径、`context_reconciliation`、证据、实际验证、延期 seam、漂移和下一路由必须可核验。实现派发额外绑定当前 Slice Implementation Contract；其他阶段不得伪造该合同。
+5. 先对照用户目标逐项验收，检查遗漏诉求、错误假设与尚未解决的问题，再验收 `Workflow Execution Result`：工作单元、合同、允许写路径、`context_reconciliation`、证据、实际验证、延期 seam、漂移和下一路由必须可核验。实现派发额外绑定当前 Slice Implementation Contract；其他阶段不得伪造该合同。
 6. 在授权边界内继续下一工作单元；遇到人工决定、新授权、证据冲突、专项失败或合并/发布结论时暂停。暂停不得阻断无依赖的独立工作。
 
 每个 `project-instance` 工作单元在申请批准或流转前，先把稳定术语回写到根目录唯一 `CONTEXT.md`，并生成通过校验的 `context_reconciliation`。候选术语、错误路径、摘要漂移或未决冲突必须阻断。质量标准只由 `engineering-baseline` 定义一次；高风险影响按裁剪合同补充反证和残余风险。
@@ -73,14 +77,16 @@ Plan 入口读取 `docs/plan/README.md` 和 `docs/process/plan-migration.md`，
 
 `机会与目标 → 业务故事 → 责任与交接 → 规则、例子与疑问 → 可验收需求 → 页面验证 → 业务任务 → 交接研发`
 
-先问清用户与结果、通常流程、责任交接、规则与例外、可独立交付的最小范围、证明结果正确的例子和证据。每轮可生成一页“业务方案总览”，只引用权威资产并显示状态、未决问题、责任人和下一步；不复制正文、不成为新的事实源、不新增门禁。
+按当前决策澄清目标、流程、责任、规则、最小范围和验收例子。可生成引用权威资产的“业务方案总览”，展示状态、未决项、责任人和下一步；不复制正文或新增门禁。
 
 Plan → Spec（含正式草稿、恢复与显式 `to-spec`）写入前，按 `docs/plan/entry-review.md` 持久化审阅包并运行 `node scripts/verify-plan-spec-entry <state.yaml>`。检查默认 pending，缺项、过期或无真实回复即阻断；独立调研可继续。
 
 ## 结果与暂停
 
-输出至少包含：模式、仓库身份、当前阶段和工作单元、影响面、资产与门禁状态、证据、新鲜度、Ticket/垂直切片/合同状态、`ready-for-agent` 结论、阻塞项、本轮动作、下一工作单元、暂停或继续理由、Ticket 同步和 Git checkpoint 判断。
+普通咨询直接回答，不创建流程产物；复杂请求简述目标与下一步。实际工作单元仍记录完整结果，按适用边界同步，对外仅展示当前决定所需信息。
+
+工作单元结果至少包含：模式、仓库身份、当前阶段和工作单元、影响面、资产与门禁状态、证据、新鲜度、Ticket/垂直切片/合同状态、`ready-for-agent` 结论、阻塞项、本轮动作、下一工作单元、暂停或继续理由、Ticket 同步和 Git checkpoint 判断。
 
 暂停会签时追加门禁 ID、`role_id`、`runtime_id`、会签文件路径、推荐答案和恢复动作；恢复前运行适用验证器。只有同一候选快照通过全部适用审查轴与 fresh verification，才能提出合并、发布或完成结论；发布仍须生物人决定。
 
-专项加载提示：实现仓库与脚手架查询 `implementation_repository_preparation`、`backend_scaffold`；切片实现查询 `ticket_formalization`、`ready_for_agent`；UI 查询 `frontend_implementation_plan`、`frontend_implementation_verification`；审查查询 `review_input`；发布查询 `release_readiness`、`git_authorization`、`user_decision_evidence`；战略交接读取 `docs/process/strategic-handoff-package.md` 并按其中验证器执行。
+专项合同加载索引见 `references/orchestration.md`。

package/template/.cursor/skills/yss-product-lifecycle/references/artifact-dependencies.md

@@ -11,10 +11,10 @@ Spec + Product Overview → System / Data Architecture
 System / Data Architecture + confirmed project architecture → Technical Design Contract (DDD / MVC)
 Technical Design Contract → Architecture Review / Engineering Baseline → implementation repository choice
 implementation repository choice → existing repository onboarding OR approved Project Scaffold Contract → scaffold generation / verification
-Repository Onboarding Result OR Scaffold Manifest + Verification → Implementation Repository Preparation Result → gate.implementation-repositories-ready
+Repository Onboarding Result OR Scaffold Manifest + Verification → Implementation Repository Preparation Result → check.implementation-repositories-ready
 API impact = yes: Requirement Freeze + Design Review + Draft Review → OpenAPI Freeze
 API impact = no: API Impact Assessment approved → No-API-Impact record；Draft/Freeze gates = not-applicable
-Frozen Contract + gate.implementation-repositories-ready → Vertical Slice Tickets
+Frozen Contract + check.implementation-repositories-ready → Vertical Slice Tickets
 Tickets + Architecture → Implementation Contract / Build Architecture Checklist
 Implementation → Independent Review → Fresh Verification → Release
 ```

package/template/.cursor/skills/yss-product-lifecycle/references/orchestration-contract.yaml

@@ -48,6 +48,63 @@ modes:
   orchestrate: bounded_write
   resume: rebuild_then_bounded_write
 
+# 用户请求理解与动作选择的唯一判断合同；自然语言解释由主控完成。
+request_triage:
+  owner: yss-product-lifecycle
+  interpretation:
+    inputs: [current-request, conversation-context, accessible-evidence]
+    capture: [goal, target, expected-deliverable, action-scope, facts, assumptions, unknowns]
+    multiple_intents: preserve-all-goals-and-order-dependencies
+    competing_interpretations: compare-effect-on-next-action
+    numeric_self_confidence: forbidden-as-routing-evidence
+  clarification:
+    trigger: missing-information-changes-next-action-or-authorization-scope
+    before_asking: reuse-context-and-inspect-accessible-evidence
+    short_request_alone: insufficient-reason-to-ask
+    question: explain-gap-and-ask-minimum-actionable-information
+    batch: independent-current-decision-questions
+    same_safe_next_step: investigate-before-asking
+    user_cannot_answer: narrow-problem-or-offer-bounded-guidance-keep-unresolved
+    timeout_is_answer: false
+  mode_selection:
+    priority: [explicit-current-restriction, authorized-continuation, clear-action-request, read-only-fallback]
+    explicit_read_only: route
+    explicit_audit: audit
+    authorized_continuation: resume
+    clear_action_request: orchestrate
+    unresolved_intent: route
+    investigation: read-only-actions-unless-additional-scope-authorized
+    mode_keywords_required: false
+    continuation: revalidate-scope-evidence-and-applicable-gates
+    gates_and_external_authorization: unchanged
+  next_action:
+    consultation: answer
+    investigation: inspect-evidence
+    action_request: execute-first-authorized-unblocked-work-unit
+    material_ambiguity: clarify
+    missing_prerequisite: explain-blocker-and-continue-independent-authorized-work
+    skill_name_only_is_completion: false
+  correction:
+    stop: actions-dependent-on-invalid-interpretation
+    preserve: still-valid-evidence-and-unrelated-user-work
+    report: revised-goal-and-actions-already-taken
+    existing_edits: explain-scope-and-agree-disposition-no-blind-rollback
+    recompute: [intent, scope, impacts, freshness, next-action]
+  acceptance:
+    per_goal: [requested-outcome, readable-evidence, unresolved-items]
+    mismatch: investigate-or-reroute
+    unresolved_triage_is_complete: false
+  presentation:
+    simple: direct-answer-or-action
+    complex: brief-goal-and-next-action
+    material_ambiguity: targeted-question
+    full_workflow_record: applicable-work-units-only
+  evaluation:
+    scenarios_ref: .agents/skills/yss-product-lifecycle/references/request-triage-scenarios.yaml
+    dimensions: [understanding-error, wrong-route, unnecessary-question, omitted-intent, post-triage-stall, correction-failure]
+    baseline: required-before-numeric-improvement-claim
+    structural_checks_prove_model_accuracy: false
+
 execution_efficiency:
   context_query:
     strategy: combine-mode-stage-work-unit-and-required-includes
@@ -259,7 +316,7 @@ work_unit_routes:
     supporting_skills: [domain-modeling, yss-research, grilling]
     skills: [yss-stage-decision, domain-modeling, yss-research, grilling]
     research_contract: {profile: strategy-evidence, mode_before_gate: evidence-audited, artifact_owner: yss-research, downstream_owner: yss-stage-decision}
-    human_gate: domain-strategy-approved
+    internal_review: check.domain-strategy-approved
     applies_when: domain_boundary_or_language_or_core_rule_impact
     not_applicable_reason: no_domain_strategy_impact
   work-unit.stage-decision:
@@ -268,7 +325,7 @@ work_unit_routes:
     supporting_skills: [domain-modeling, yss-research]
     skills: [yss-stage-decision, domain-modeling, yss-research]
     research_contract: {profile: strategy-evidence, mode_before_gate: evidence-audited, artifact_owner: yss-research, downstream_owner: yss-stage-decision}
-    human_gate: stage-decision-package-approved
+    internal_review: check.stage-decision-package-approved
     applies_when: plan_to_spec_requires_stable_decision_package
     not_applicable_reason: existing_approved_stage_decision_current
   work-unit.spec-synthesis:
@@ -297,7 +354,7 @@ work_unit_routes:
         H2: {kind: flow-review, adapter: runnable-flow, react_antd: supported-not-required, antd_facts: conditional}
     implementation_handoff: {production_component_assumptions: required-array, verification_targets: frontend-implementation-only}
     version_boundary: {h2_design_standard: vue-antdv-next-default-react-antd-explicit-compatibility, prototype_must_not_call: yss-ui, production_version_source: implementation-lockfile}
-    human_gate: prototype-confirmation
+    human_gate: gate.product-design-approved
     applies_when: product_design_impact
     not_applicable_reason: no_ui_or_product_design_impact
   work-unit.technical-analysis:
@@ -319,7 +376,7 @@ work_unit_routes:
       statuses: [draft, ready-for-human, approved, blocked, stale, drift, new_impacts, not-applicable]
       approved_by: yss-product-lifecycle
       forbidden: [self-approval, ready-for-agent, production-code-generation]
-    human_gate: openapi-freeze_if_api_impact
+    human_gate: gate.engineering-contract-approved
     applies_when: technical_or_contract_impact
     not_applicable_reason: no_technical_or_contract_impact
   work-unit.implementation-repository-preparation:
@@ -331,7 +388,7 @@ work_unit_routes:
     not_applicable_reason: no_frontend_or_backend_delivery_impact
     decision_boundary: scaffold-choice
     completion_states: [existing-and-onboarded, initialized-and-verified, not-applicable]
-    aggregate_gate: gate.implementation-repositories-ready
+    automatic_check: check.implementation-repositories-ready
     must_complete_before: work-unit.ticket-decomposition
   work-unit.ticket-decomposition:
     native: {source: yss-product-lifecycle, skill: yss-product-lifecycle, invocation_mode: model-invoked}
@@ -433,7 +490,8 @@ work_unit_routes:
     primary_skill: yss-product-lifecycle
     supporting_skills: []
     skills: [yss-product-lifecycle]
-    human_gate: merge_or_release
+    acceptance_gate: gate.delivery-accepted
+    external_action_authorization: explicit_merge_or_release_only
     applies_when: release_or_stage_completion
     not_applicable_reason: work_not_ready_for_release
 
@@ -474,7 +532,7 @@ ticket_formalization:
 implementation_repository_preparation:
   stage: stage.system-data-engineering
   work_unit: work-unit.implementation-repository-preparation
-  gate: gate.implementation-repositories-ready
+  check: check.implementation-repositories-ready
   applies_per_delivery_role: [backend, frontend]
   affected_role_allowed_states: [existing-and-onboarded, initialized-and-verified]
   unaffected_role_state: not-applicable
@@ -856,7 +914,7 @@ lifecycle_work_unit_requirements:
 release_readiness:
   required: [fresh_verification, release_or_rollback_evidence, independent_review, all_triggered_gates_approved_or_not_applicable]
   conditional:
-    ui_impact: [gate.frontend-implementation-verified, evidence.frontend-implementation-verification]
+    ui_impact: [check.frontend-implementation-verified, evidence.frontend-implementation-verification]
   blocked_when: [triggered_gate_blocked, triggered_gate_stale, triggered_gate_not_evaluated, missing_evidence]
 
 wayfinder_completion:
@@ -1057,3 +1115,17 @@ user_decision_evidence:
   response_absent: paused-human-gate
   invalid_response: blocked
   basis_changed: stale
+
+gate_consolidation:
+  source: docs/process/lifecycle-registry.yaml
+  check_policy: docs/agents/digital-human-roles.yaml
+  checks: automatic_or_independent_review_without_separate_user_decision
+  check_failure: blocks_aggregate_gate
+  plan_decision_boundary: gate.plan-approved
+  readiness: computed_after_current_approved_contract_and_all_applicable_preconditions
+  api_freeze: atomic_with_engineering_contract_approval_same_version
+  acceptance_does_not_authorize: [merge, push, release, production_change]
+  reuse: unchanged_subject_basis_and_scope_only
+  invalidation: affected_dependencies_or_authorization_change_only
+  historical_gate_records: read_only_no_automatic_approval_migration
+  verifier: scripts/verify-lifecycle-checkpoint

package/template/.cursor/skills/yss-product-lifecycle/references/orchestration.md

@@ -2,7 +2,7 @@
 
 ## 有界推进循环
 
-1. 识别模式、仓库身份、任务规模和影响面。
+1. 按 `orchestration-contract.yaml.request_triage` 理解请求并选择模式，再识别仓库身份、任务规模和影响面；问题理解与澄清细节见 [请求分诊协议](request-triage.md)。
 2. `setup readiness`：每个任务只执行一次，核对 tracker、五态标签和领域文档布局，并在本轮缓存结果；仅在 tracker、主远端、真实标签或配置变化时重查。
 3. 加载父 Ticket/checkpoint 与真实资产，计算最近可信阶段。
 4. 评估资产、门禁和 `stale`，选择第一个未阻塞工作单元。进入 `work-unit.slice-implementation` 前，必须先通过 `scripts/lib/lifecycle-transition.mjs` 的 Ticket 正式化、垂直切片绑定和合法 `next_route` 校验；父 Ticket、缺少垂直切片或 `ready-for-human` 的切片一律 `blocked`。
@@ -39,7 +39,7 @@ Matt phase boundary 是工作阶段之间的上下文决策，不是新的生命
 
 阶段 5 在技术分析后固定进入 `work-unit.implementation-repository-preparation`。编排器按 backend / frontend 影响逐项目展示已有或新建工程、Agent 推荐、仓库 scope、目标路径、Git 初始化、验证命令和风险，取得真实用户确认后再执行。已有工程调用 `implementation-repo-onboarding`；新后端按已确认架构调用对应生成器；新前端调用 `yss-frontend-scaffold-generator`。默认 `external-repository`，`harness-apps` / `git-submodule`、`git init` 和远端仓库创建均须显式选择或授权。
 
-所有命中项目必须达到 `existing-and-onboarded` 或 `initialized-and-verified`，未命中的交付面必须有带原因的 `not-applicable`。聚合结果、Manifest、验证和 onboarding 证据必须可读且当前；否则 `gate.implementation-repositories-ready` 阻断 Ticket 正式化。旧实例恢复时保留已有 Ticket，但将相关 Ticket / Slice Contract 标为 `blocked` / `stale` 并返回本工作单元。
+所有命中项目必须达到 `existing-and-onboarded` 或 `initialized-and-verified`，未命中的交付面必须有带原因的 `not-applicable`。聚合结果、Manifest、验证和 onboarding 证据必须可读且当前；否则 `check.implementation-repositories-ready` 阻断 Ticket 正式化。旧实例恢复时保留已有 Ticket，但将相关 Ticket / Slice Contract 标为 `blocked` / `stale` 并返回本工作单元。
 
 `prototype_confirmation` 通过后，先判断实现仓库登记中的 backend `scaffold_status`。当状态为 `required` 时，在工程基线中先完成脚手架架构选择：Agent 按领域复杂度给出 `domain-driven` / `layered-mvc` 推荐和依据，本体选择作为子项目预填默认值，用户通过批量表确认全部项目或逐项覆盖。选择写入 `scaffold-architecture-decisions.yaml`；处于 `undecided`、`recommended`、`awaiting-user-decision` 或 `stale` 时必须阻断，不得默认 DDD，也不得在生成器内交互。
 
@@ -82,7 +82,7 @@ tracker 选择和冲突按 `docs/agents/issue-tracker.md` 裁决：已持久化
 - 小改动和中等变更可由同一独立执行者完成 `code-review` 与 fresh verification，并在同一报告中分别记录 findings、命令、结果和残余风险。
 - 该执行者必须独立于实现者；新模块、高风险变更、职责冲突或需双人控制时，Reviewer 与 Verifier 分开。
 - `code-review` 是唯一默认代码审查 skill。GitLab、CI、Sonar、Alibaba Java 与 YSS 前端 / 后端 skill 作为仓库规则或专项检查输入接入 Standards 轴，由 `review_standards_route` 按影响面编译；不再叠加第二个通用审查 skill。漏掉合同 `required_skills`、适用报告行空白、mandatory `violation` 未关闭或可机器检查规则既无工具结果也无原文引用时，不得 `completed`。
-- 产品切片与模板维护共用同一 finding 闭环，强度分别绑定 Slice 合同与 L1 / L2 / L3。`violation`、机器检查失败、适用行空白由实现者在原合同路径修复，再重新捕获候选并全轴复审。`drift`、`new_impacts`、`required_skills` 与真实影响不一致时合同 `stale`，回 实现合同编译器 或更早阶段，禁止在旧合同上继续编码。审查者不得写实现。`not-applicable` 仅当影响面未命中；命中后 mandatory 不得豁免，只允许修复或完整 `seam-deferred`。禁止为日常 Alibaba / YSS 新增生物人豁免门禁。
+- 产品切片与模板维护共用 finding 分类；产品切片执行下述候选审查闭环，模板维护按 `docs/process/harness-process-tailoring.md` 的 L1 / L2 / L3 验证强度闭合，不自动冻结候选或全轴复审。`violation`、机器检查失败、适用行空白由实现者在原合同路径修复，再重新捕获候选并全轴复审。`drift`、`new_impacts`、`required_skills` 与真实影响不一致时合同 `stale`，回 实现合同编译器 或更早阶段，禁止在旧合同上继续编码。审查者不得写实现。`not-applicable` 仅当影响面未命中；命中后 mandatory 不得豁免，只允许修复或完整 `seam-deferred`。禁止为日常 Alibaba / YSS 新增生物人豁免门禁。
 - 独立 Reviewer 必须与实现者不同实例，并在能执行已登记 `pnpm` / `./mvnw` 的运行时中审查。模板源 `.cursor/environment.json` 只服务模板校验，不替代实现仓审查运行时。不为此再创建第二个 Cloud 审查环境或 `/code-review` skill。
 - UI 影响切片将 `UI fidelity` 作为 `code-review` 的条件第三轴；任何修复都会使候选摘要失效，必须重新捕获候选并重跑 Standards、Spec、UI fidelity 和 fresh verification。
 
@@ -98,7 +98,7 @@ tracker 选择和冲突按 `docs/agents/issue-tracker.md` 裁决：已持久化
 
 ## 必须暂停
 
-- Spec baseline、需求冻结、原型确认、OpenAPI Freeze 或 Architecture Review 等普通门禁等待会签裁决（数字人或生物人，以 `docs/agents/digital-human-roles.yaml` 的 `gate_policy` 为准）。暂停输出必须包含：门禁 ID、指定 `role_id`、`runtime_id`、会签文件路径。恢复前执行 `scripts/verify-approval-record --require-approved`；角色错误、起草者自签或生物人门禁被数字人关闭时返回 `blocked`，不得标 `approved`。
+- 注册表中的聚合门禁等待会签裁决（数字人或生物人，以 `docs/agents/digital-human-roles.yaml` 的 `gate_policy` 为准）。暂停输出必须包含：门禁 ID、指定 `role_id`、`runtime_id`、会签文件路径。恢复前执行 `scripts/verify-approval-record --require-approved`；角色错误、起草者自签或生物人门禁被数字人关闭时返回 `blocked`，不得标 `approved`。
 - 需要目标仓库、外部凭据、发布窗口或其他新授权。
 - 状态与证据冲突且无法可靠重建。
 - 专项 skill 失败或返回不可验收结果。
@@ -134,3 +134,17 @@ Matt `prototype` 的回流还必须注明 `prototype_branch`，并保留单文
 `to-questionnaire` 未收到答案时使用 `external-input-required` 暂停，记录问卷、接收人、所需输出和恢复路由；收到答案后记录 response、重新分类影响面和更新后的权威资产，再回到 `grill-with-docs` 或 `to-spec`。
 
 Release 与 Retrospective 属于生命周期编排器拥有的工作单元。发布和复盘前都必须重新取得 fresh verification；发布还需要发布/回滚证据和独立审查，复盘还需要复盘记录和治理回流判断，再回流权威资产。
+
+## 聚合门禁和检查证据
+
+`gates` 仅记录正式批准 / 验收；`checks` 记录内部专业审查、自动前置核验和就绪计算。检查职责读取角色表 `check_reviews` / `automatic_checks`，命中项失败仍阻断。Plan 内部审查使用入口审阅包的 `internal_checks`，统一以 `gate.plan-approved` 获取用户决定。工程契约审查通过并批准当前版本后同时冻结 OpenAPI，冻结准备不单独索取批准。
+
+Checkpoint 中已批准门禁必须带 `basis: [{ref, digest}]` 和按 `evidence.*` 分类的 `evidence` 引用数组；此处 digest 是实际文件字节的 64 位十六进制 SHA-256。每个适用检查带 `applicable: true`、通过状态和同样的证据绑定；不适用项须 `applicable: false`、原因和影响面证据。内部专业审查还须 `approval_ref`、`subject_ref`、`approval_scope`，审查记录带 `subject_digest` 和不同于 `principal_ref` 的 `drafter_principal_ref`。
+
+聚合门禁的 `subject_ref` 指向当前审阅包，包含 `gate_id` 和 `basis`，覆盖所有检查及其他验收依据；用户批准绑定此包，专业会签记录绑定其 `subject_digest`。检查证据、审阅包和会签记录均纳入 checkpoint 的摘要绑定，避免只改变检查结果却复用旧批准。`scripts/verify-lifecycle-checkpoint` 实际核验；旧 gate ID 只允许历史读取或按源策略校验外部冻结包，不能直接关闭当前门禁。
+
+交付验收可由测试角色完成；实际合并、推送和发布分别执行已有外部动作授权检查。阶段完成不请求发布权限。没有相关资产、依据或范围变化时复用现有批准；相关变化仅失效受影响批准。
+
+## 专项合同加载索引
+
+专项加载提示：实现仓库与脚手架查询 `implementation_repository_preparation`、`backend_scaffold`；切片实现查询 `ticket_formalization`、`ready_for_agent`；UI 查询 `frontend_implementation_plan`、`frontend_implementation_verification`；审查查询 `review_input`；发布查询 `release_readiness`、`git_authorization`、`user_decision_evidence`；战略交接读取 `docs/process/strategic-handoff-package.md` 并按其中验证器执行。