dsh-gate-agent 0.1.1 → 0.1.2

package/skills/gate-verify/SKILL.md

@@ -1,73 +1,73 @@
----
-name: gate-verify
-description: Gate 验收方法：主 agent 读取后提炼并派发；worker 不加载技能。
-user-invocable: false
----
-
-## Main agent use
-
-主 agent 在派发前自行读取本方法，给 verify worker：工作项、验收条件、受保护基线和验证规则。worker 不应加载本技能或其他 gate 方法文件。
-
-任务模板：本技能 `assets/worker-briefs/gate-verify.md`。
-
-
-
-## 验收
-
-用证据确认授权工作满足其权威合同，完成工程闭环。验收包含授权范围内的修正与再验证。
-
-## 1. 建立验收合同与覆盖
-
-确定工作项及其批准要求、验收条件、设计约束和兼容性预期。验收对象是实现，不能把实现当作期望行为的来源。验证保持在已同意合同内。
-
-对每个必需验收条件，选择能证明它的最高稳定边界，例如用户可见行为、公开 API/CLI、集成或领域边界。使用最小充分检查；更低层测试、构建和静态检查可在适当处支持证据。
-
-已有检查能证明合同时复用。测试通过、CI 或 code review 本身不等于验收通过；不要因为检查存在就运行。
-
-## 2. 执行验证
-
-条件允许时实际执行交付行为。逐项记录必需条件、证据和当前结果：
-
-- **PASS**：证据证明必需行为；
-- **FAIL**：观察行为与必需行为矛盾；
-- **BLOCKED**：歧义或可靠证据不足，无法判定。
-
-没有观察到失败不等于 PASS。发现失败或 blocker 后仍继续检查相互独立的条件。改动可能影响既有相邻行为时，按影响和风险比例做回归检查。
-
-## 3. 诊断失败并交回
-
-对每个失败，确定合同期望、触发条件、观察行为和支撑证据。改变任何东西前确认原因和属主；未解决原因要明确记录，不能猜：
-
-- **实现缺陷**：报告给 implement worker；不要在这里修。
-- **验证缺陷**：把错误检查报告给 gate；不要在这里改检查。
-- **合同歧义或缺陷**：停止受影响验证并报告；gate 把合同退回 planning。
-- **环境或工具失败**：报告什么阻止继续验证，以及是否在范围内。
-- **既有失败**：记录其对验收的影响和属主；不要自动扩大修复范围，也不要忽略它阻止通过的必需条件。
-
-已证明的合同违反即使根因未解决也保持 FAIL。
-
-gate 把修正路由给属主 worker，然后再次验证。授权合同变更改变验收基线时，重做 §1。
-
-不要为了通过验证而削弱已同意合同。
-
-## 4. 最终结果与完成
-
-先给最终验收结果，再按影响排序未解决失败和 blocker。随后给每个必需条件的简洁覆盖和证据、已完成修正及再验证证据、相关回归结果、实质验证缺口或残余风险。阻止可靠判定的缺口必须反映到对应条件的 BLOCKED。
-
-报告最新已验证状态：修正后通过再验证的条件为 PASS，并把它此前的失败概括为已完成修正。未解决事项按影响排序，写明合同引用、期望与观察行为、原因或不确定性、所需下一步。
-
-按当前必需条件确定总体结果：
-
-- **PASS**：每个必需条件都通过；
-- **FAIL**：任一必需条件仍失败，即使其他条件 blocked；
-- **BLOCKED**：没有失败，但一项或多项无法可靠验证。
-
-### 完成
-
-报告 PASS/FAIL/BLOCKED 和证据。不要更新工作项、转换决策记录或完成 gate；这些由 gate 接受报告后处理。
-
-
-
-
+---
+name: gate-verify
+description: Gate 验收方法：主 agent 读取后提炼并派发；worker 不加载技能。
+user-invocable: false
+---
+
+## Main agent use
+
+主 agent 在派发前自行读取本方法，给 verify worker：工作项、验收条件、受保护基线和验证规则。worker 不应加载本技能或其他 gate 方法文件。
+
+任务模板：本技能 `assets/worker-briefs/gate-verify.md`。
+
+
+
+## 验收
+
+用证据确认授权工作满足其权威合同，完成工程闭环。验收包含授权范围内的修正与再验证。
+
+## 1. 建立验收合同与覆盖
+
+确定工作项及其批准要求、验收条件、设计约束和兼容性预期。验收对象是实现，不能把实现当作期望行为的来源。验证保持在已同意合同内。
+
+对每个必需验收条件，选择能证明它的最高稳定边界，例如用户可见行为、公开 API/CLI、集成或领域边界。使用最小充分检查；更低层测试、构建和静态检查可在适当处支持证据。
+
+已有检查能证明合同时复用。测试通过、CI 或 code review 本身不等于验收通过；不要因为检查存在就运行。
+
+## 2. 执行验证
+
+条件允许时实际执行交付行为。逐项记录必需条件、证据和当前结果：
+
+- **PASS**：证据证明必需行为；
+- **FAIL**：观察行为与必需行为矛盾；
+- **BLOCKED**：歧义或可靠证据不足，无法判定。
+
+没有观察到失败不等于 PASS。发现失败或 blocker 后仍继续检查相互独立的条件。改动可能影响既有相邻行为时，按影响和风险比例做回归检查。
+
+## 3. 诊断失败并交回
+
+对每个失败，确定合同期望、触发条件、观察行为和支撑证据。改变任何东西前确认原因和属主；未解决原因要明确记录，不能猜：
+
+- **实现缺陷**：报告给 implement worker；不要在这里修。
+- **验证缺陷**：把错误检查报告给 gate；不要在这里改检查。
+- **合同歧义或缺陷**：停止受影响验证并报告；gate 把合同退回 planning。
+- **环境或工具失败**：报告什么阻止继续验证，以及是否在范围内。
+- **既有失败**：记录其对验收的影响和属主；不要自动扩大修复范围，也不要忽略它阻止通过的必需条件。
+
+已证明的合同违反即使根因未解决也保持 FAIL。
+
+gate 把修正路由给属主 worker，然后再次验证。授权合同变更改变验收基线时，重做 §1。
+
+不要为了通过验证而削弱已同意合同。
+
+## 4. 最终结果与完成
+
+先给最终验收结果，再按影响排序未解决失败和 blocker。随后给每个必需条件的简洁覆盖和证据、已完成修正及再验证证据、相关回归结果、实质验证缺口或残余风险。阻止可靠判定的缺口必须反映到对应条件的 BLOCKED。
+
+报告最新已验证状态：修正后通过再验证的条件为 PASS，并把它此前的失败概括为已完成修正。未解决事项按影响排序，写明合同引用、期望与观察行为、原因或不确定性、所需下一步。
+
+按当前必需条件确定总体结果：
+
+- **PASS**：每个必需条件都通过；
+- **FAIL**：任一必需条件仍失败，即使其他条件 blocked；
+- **BLOCKED**：没有失败，但一项或多项无法可靠验证。
+
+### 完成
+
+报告 PASS/FAIL/BLOCKED 和证据。不要更新工作项、转换决策记录或完成 gate；这些由 gate 接受报告后处理。
+
+
+
+
 
 

package/skills/gate-verify/assets/worker-briefs/gate-verify.md

@@ -1,21 +1,21 @@
-# gate-verify worker
-
-```markdown
-你是 Gate 的 verify worker。使用 gate-verify 的验收方法确认交付满足权威合同。
-
-## 验收对象
-<工作项、交付范围、固定点>
-
-## 验收合同
-<需求、验收条件、兼容性/设计约束>
-
-## 证据要求
-- 用可观察 seam 验证，不只依赖测试通过。
-- 记录实际执行的命令、证据文件和结果。
-- 无法可靠判断时报告失败或阻塞，不猜 PASS。
-
-## 输出
-返回验收结果、失败项、证据、回归检查和剩余风险；同时返回你实际使用的 `call_...` tool result id 列表，供 root 填 `verification.toolResults`。不做无关修复。
-```
-
-
+# gate-verify worker
+
+```markdown
+你是 Gate 的 verify worker。使用 gate-verify 的验收方法确认交付满足权威合同。
+
+## 验收对象
+<工作项、交付范围、固定点>
+
+## 验收合同
+<需求、验收条件、兼容性/设计约束>
+
+## 证据要求
+- 用可观察 seam 验证，不只依赖测试通过。
+- 记录实际执行的命令、证据文件和结果。
+- 无法可靠判断时报告失败或阻塞，不猜 PASS。
+
+## 输出
+返回验收结果、失败项、证据、回归检查和剩余风险；同时返回你实际使用的 `call_...` tool result id 列表，供 root 填 `verification.toolResults`。不做无关修复。
+```
+
+

package/skills/gate-write-agent-notes/SKILL.md

@@ -160,7 +160,7 @@ Status: <状态>
 
 这个启动器只调用插件内置脚本，不复制技能。
 
-- 计划完成前或提交前，运行 `npm run verify-notes`，检查 tree、format、lifecycle、archived。
+- 计划完成前或提交前，运行 `npm run verify-notes`，检查 tree、format、lifecycle、archived。
 - `/gate-init` 同时生成 `.github/workflows/verify-notes.yml` 和 `.agents/gate/ci/` 校验脚本副本；CI 使用项目内副本，不依赖插件安装路径。
 - gate 会话内也可以用 `/gate-note-verify` 做同样的只读检查。
 - Verify PASS 后由 gate 执行 `gate_note_finalize` → `gate_note_verify` 工具 → `gate_record({node:'note-lifecycle'})`；lifecycle 不要用 npm 代替。

package/skills/gate-write-agent-notes/scripts/verify-agent-note-lifecycle.ts

@@ -1,74 +1,74 @@
-/**
- * Lifecycle check for Agent Notes: one topic, one active lifecycle.
- *
- * This is the generic part of the rule; it does not read the capability ledger.
- * A project may add a wrapper for product-specific status.
- */
-import { readdirSync, readFileSync } from 'node:fs'
-import path from 'node:path'
-
-const ROOT = process.argv.includes('--root')
-  ? process.argv[process.argv.indexOf('--root') + 1]
-  : process.argv.find(arg => arg.startsWith('--root='))?.slice('--root='.length) ?? '.agents/notes'
-const LIFEcycles = ['proposed', 'implemented', 'rejected']
-const STATUS = { proposed: 'proposed', implemented: 'implemented', rejected: 'rejected' }
-const errors = []
-const byTopic = new Map()
-
-function normalize(value) {
-  return String(value).toLowerCase().replace(/[^a-z0-9]+/gu, '-').replace(/^-|-$/gu, '')
-}
-
-function topicOf(file) {
-  const base = path.basename(file, '.md')
-  return normalize(base.replace(/^\d{4}-\d{2}-\d{2}-/u, '').replace(/-implemented$/u, ''))
-}
-
-function walk(lifecycle) {
-  const dir = path.join(ROOT, lifecycle)
-  const files = []
-  for (const entry of readdirSync(dir, { withFileTypes: true })) {
-    if (entry.isDirectory()) {
-      const sub = path.join(dir, entry.name)
-      for (const nested of readdirSync(sub, { withFileTypes: true })) {
-        if (nested.isFile() && nested.name.endsWith('.md')) files.push(path.join(sub, nested.name))
-      }
-    } else if (entry.isFile() && entry.name.endsWith('.md')) {
-      files.push(path.join(dir, entry.name))
-    }
-  }
-  return files
-}
-
-for (const lifecycle of LIFEcycles) {
-  for (const file of walk(lifecycle)) {
-    const topic = topicOf(file)
-    if (topic === '') {
-      errors.push(`${file}: 无法从文件名得到 topic`)
-      continue
-    }
-    const status = /^Status:\s*(\S+)/mu.exec(readFileSync(file, 'utf8'))?.[1]
-    if (status !== STATUS[lifecycle]) {
-      errors.push(`${file}: Status=${String(status)}，目录要求 ${STATUS[lifecycle]}`)
-    }
-    const entry = byTopic.get(topic) ?? []
-    entry.push({ lifecycle, file })
-    byTopic.set(topic, entry)
-  }
-}
-
-for (const [topic, entries] of byTopic) {
-  if (entries.length > 1) {
-    errors.push(`${topic}: 同主题多份（${entries.map(entry => `${entry.lifecycle}:${path.basename(entry.file)}`).join('；')}）`)
-  }
-  const lifecycles = new Set(entries.map(entry => entry.lifecycle))
-  if (lifecycles.has('proposed') && lifecycles.has('implemented')) {
-    errors.push(`${topic}: proposed 与 implemented 不能并存`)
-  }
-}
-
-if (errors.length > 0) {
-  for (const error of errors) console.error(`FAIL ${error}`)
-  process.exit(1)
-}
-console.log(`ok: ${byTopic.size} note topics, one lifecycle each`)
+/**
+ * Lifecycle check for Agent Notes: one topic, one active lifecycle.
+ *
+ * This is the generic part of the rule; it does not read the capability ledger.
+ * A project may add a wrapper for product-specific status.
+ */
+import { readdirSync, readFileSync } from 'node:fs'
+import path from 'node:path'
+
+const ROOT = process.argv.includes('--root')
+  ? process.argv[process.argv.indexOf('--root') + 1]
+  : process.argv.find(arg => arg.startsWith('--root='))?.slice('--root='.length) ?? '.agents/notes'
+const LIFEcycles = ['proposed', 'implemented', 'rejected']
+const STATUS = { proposed: 'proposed', implemented: 'implemented', rejected: 'rejected' }
+const errors = []
+const byTopic = new Map()
+
+function normalize(value) {
+  return String(value).toLowerCase().replace(/[^a-z0-9]+/gu, '-').replace(/^-|-$/gu, '')
+}
+
+function topicOf(file) {
+  const base = path.basename(file, '.md')
+  return normalize(base.replace(/^\d{4}-\d{2}-\d{2}-/u, '').replace(/-implemented$/u, ''))
+}
+
+function walk(lifecycle) {
+  const dir = path.join(ROOT, lifecycle)
+  const files = []
+  for (const entry of readdirSync(dir, { withFileTypes: true })) {
+    if (entry.isDirectory()) {
+      const sub = path.join(dir, entry.name)
+      for (const nested of readdirSync(sub, { withFileTypes: true })) {
+        if (nested.isFile() && nested.name.endsWith('.md')) files.push(path.join(sub, nested.name))
+      }
+    } else if (entry.isFile() && entry.name.endsWith('.md')) {
+      files.push(path.join(dir, entry.name))
+    }
+  }
+  return files
+}
+
+for (const lifecycle of LIFEcycles) {
+  for (const file of walk(lifecycle)) {
+    const topic = topicOf(file)
+    if (topic === '') {
+      errors.push(`${file}: 无法从文件名得到 topic`)
+      continue
+    }
+    const status = /^Status:\s*(\S+)/mu.exec(readFileSync(file, 'utf8'))?.[1]
+    if (status !== STATUS[lifecycle]) {
+      errors.push(`${file}: Status=${String(status)}，目录要求 ${STATUS[lifecycle]}`)
+    }
+    const entry = byTopic.get(topic) ?? []
+    entry.push({ lifecycle, file })
+    byTopic.set(topic, entry)
+  }
+}
+
+for (const [topic, entries] of byTopic) {
+  if (entries.length > 1) {
+    errors.push(`${topic}: 同主题多份（${entries.map(entry => `${entry.lifecycle}:${path.basename(entry.file)}`).join('；')}）`)
+  }
+  const lifecycles = new Set(entries.map(entry => entry.lifecycle))
+  if (lifecycles.has('proposed') && lifecycles.has('implemented')) {
+    errors.push(`${topic}: proposed 与 implemented 不能并存`)
+  }
+}
+
+if (errors.length > 0) {
+  for (const error of errors) console.error(`FAIL ${error}`)
+  process.exit(1)
+}
+console.log(`ok: ${byTopic.size} note topics, one lifecycle each`)

package/skills/gate-writing-for-agents/SKILL-MECHANICS.md

@@ -6,12 +6,12 @@ The skill-specific branch of [`gate-writing-for-agents`](SKILL.md): what changes
 
 Two choices, trading the two loads:
 
-- A **model-invoked** skill keeps a `description`, so the agent can fire it autonomously — and other skills can reach it. You can still type its name: model-invocation always _includes_ user reach; a description only ever adds agent discovery, never removes the human's. The description is the skill's top-level context pointer, forced to stay loaded at all times — permanent context load in exchange for discoverability. A model-invoked skill whose content is all reference is also one home for shared reference: another skill can invoke it, so reference needed by several skills lives in one place. Mechanics: omit `policy.allow_implicit_invocation` from `agents/openai.yaml` (or leave it enabled), and write a model-facing description carrying the trigger branches (the pointer-writing rules in `SKILL.md` apply in full).
-- A **user-invoked** skill is not eligible for automatic activation: only the human typing its name can invoke it. It spends cognitive load — you are the index that must remember it exists. Mechanics: set `policy.allow_implicit_invocation: false` in `agents/openai.yaml`; keep the `description` as a concise human-facing summary with trigger lists stripped.
+- A **model-invoked** skill keeps a `description`, so the agent can fire it autonomously — and other skills can reach it. You can still type its name: model-invocation always _includes_ user reach; a description only ever adds agent discovery, never removes the human's. The description is the skill's top-level context pointer, forced to stay loaded at all times — permanent context load in exchange for discoverability. A model-invoked skill whose content is all reference is also one home for shared reference: another skill can invoke it, so reference needed by several skills lives in one place. Mechanics: omit `policy.allow_implicit_invocation` from `agents/openai.yaml` (or leave it enabled), and write a model-facing description carrying the trigger branches (the pointer-writing rules in `SKILL.md` apply in full).
+- A **user-invoked** skill is not eligible for automatic activation: only the human typing its name can invoke it. It spends cognitive load — you are the index that must remember it exists. Mechanics: set `policy.allow_implicit_invocation: false` in `agents/openai.yaml`; keep the `description` as a concise human-facing summary with trigger lists stripped.
 
 Pick model-invocation only when the agent must reach the skill on its own, or another skill must. If it only ever fires by hand, make it user-invoked and pay no context load.
 
-Shared reference that two user-invoked skills both need can live in neither because neither is eligible for automatic activation. Push it to a plain file outside the skill system: external reference any skill can point at.
+Shared reference that two user-invoked skills both need can live in neither because neither is eligible for automatic activation. Push it to a plain file outside the skill system: external reference any skill can point at.
 
 ## Splitting by invocation
 
@@ -19,6 +19,6 @@ The invocation cut of splitting (the sequence cut lives in `SKILL.md`): split of
 
 ## Router skills
 
-When user-invoked skills multiply past what you can remember, that piled-up cognitive load is cured by a **router skill**: one user-invoked skill that names the others and when to reach for each, so the human has one skill to remember instead of many. It can only hint, never fire them: user-invoked skills are not eligible for automatic activation, so nothing but the human can reach them.
+When user-invoked skills multiply past what you can remember, that piled-up cognitive load is cured by a **router skill**: one user-invoked skill that names the others and when to reach for each, so the human has one skill to remember instead of many. It can only hint, never fire them: user-invoked skills are not eligible for automatic activation, so nothing but the human can reach them.