feat(cs): dispatch clear action requests in the same turn

Dogfood feedback (session 001c553b): calling cs with a concrete request
recommended an entry and then waited for the user to re-invoke it — one
confirmation too many. cs now classifies intent: clear action requests
announce the chosen entry in one line and continue executing under that
skill's discipline in the same turn; advice requests still get a
recommendation only; ambiguous requests that would materially change what
happens next get exactly one focused question. Handoff does not widen
authority — the target skill's gates and checkpoints apply unchanged.
Safety anchors and docs (SKILL_CATALOG / WORKFLOW / README zh+en) synced.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
dafang
2026-07-30 20:44:37 +08:00
parent a9eaa93ff8
commit 7a40410cfe
8 changed files with 26 additions and 18 deletions
+1 -1
View File
@@ -182,7 +182,7 @@ High-risk changes, or an explicit owner request, invoke `cs-code-review` for ind
| Group | Skill | Purpose |
|---|---|---|
| Navigation | `cs` | Explain the system and recommend an entry; never start another workflow or write files |
| Navigation | `cs` | Clear action requests dispatch to the target skill in the same turn; advice gets a recommendation only; the overview writes no files |
| Onboard | `cs-onboard` | Install CodeStable into a repository |
| Epic | `cs-epic` | Decompose, confirm, and drive multiple deliverable items over time |
| Feature | `cs-feat` | Implement new capability and scale design confirmation or review with risk |
+1 -1
View File
@@ -185,7 +185,7 @@ CodeStable 顺着软件编码的真实流程来设计,把开发活动建模成
| 分组 | 技能 | 用途 |
|---|---|---|
| 导航 | `cs` | 解释体系并推荐入口;不启动下游流程、不写文件 |
| 导航 | `cs` | 明确行动诉求同轮直转对应入口;咨询只推荐;导览不写文件 |
| 接入 | `cs-onboard` | 把 CodeStable 接入新仓库或已有零散文档仓库 |
| 大需求 | `cs-epic` | 拆解、确认并长程推进多个子项 |
| 功能 | `cs-feat` | 实现新功能或功能改造,按风险升级设计确认与 review |
+1 -1
View File
@@ -7,7 +7,7 @@ skill files or a centralized onboard runtime.
| Group | Skill | Responsibility |
|---|---|---|
| Navigation | `cs` | Explain CodeStable and recommend one entry; never start a workflow or write files |
| Navigation | `cs` | Clear action requests dispatch to the target skill in the same turn; advice gets a recommendation only; the overview writes no files |
| Onboarding | `cs-onboard` | Create minimal project memory and explain a lossless v1 upgrade |
| Feature | `cs-feat` | Implement new capability; scale design confirmation and independent review with risk |
| Issue | `cs-issue` | Fix bugs or broken existing behavior with red-to-green evidence |
+1 -1
View File
@@ -7,7 +7,7 @@ runtime。
| 分组 | Skill | 责任 |
|---|---|---|
| 导航 | `cs` | 解释 CodeStable 并推荐一个入口;不启动流程、不写文件 |
| 导航 | `cs` | 明确行动诉求同轮直转对应入口;咨询只推荐;导览不写文件 |
| 接入 | `cs-onboard` | 创建最小项目记忆骨架;无损说明 v1 升级边界 |
| 功能 | `cs-feat` | 实现新功能;按风险决定是否先确认设计或做独立 review |
| 问题 | `cs-issue` | 用红到绿的验证修复 bug 或既有行为异常 |
+3 -2
View File
@@ -3,8 +3,9 @@
## Workflow
CodeStable v2 consists of eight independently installed thin-harness skills and a project-memory
loop. `cs` only explains and recommends; it does not start another workflow automatically. Once the
entry is known, invoke that skill directly.
loop. `cs` classifies what the user wants right now: clear action requests dispatch to the target
skill in the same turn and continue executing; advice requests get a recommendation only; with no
request it presents the overview.
```text
unsure which entry -> cs
+2 -2
View File
@@ -2,8 +2,8 @@
## 工作流
CodeStable v2 是 8 个独立安装的 thin-harness skill加一个项目记忆闭环。`cs` 只解释和
推荐,不自动启动下游流程;确定入口后直接调用对应 skill
CodeStable v2 是 8 个独立安装的 thin-harness skill加一个项目记忆闭环。`cs` 判别用户
此刻要什么:明确行动诉求**同轮直转**对应 skill 并继续执行;咨询只给推荐;无诉求时介绍体系
```text
不确定入口 -> cs
+16 -9
View File
@@ -1,16 +1,23 @@
---
name: cs
description: CodeStable 体系速读。触发:用户想了解 CodeStable、问该用哪个 cs skill不确定诉求归属
argument-hint: "[问题]"
description: CodeStable 入口。触发:用户调用 cs、想了解体系、问该用哪个 skill带着诉求未选入口。明确行动诉求同轮直转对应 skill
argument-hint: "[诉求]"
---
# cs
介绍 CodeStable 并帮用户选对入口。本 skill 只解释和推荐,不启动流程、不写文件。
判断用户此刻要什么:要么直接开始干,要么给推荐,要么介绍体系。**明确的行动诉求不停下等确认。**
## 体系速读
## 判别与行为
CodeStable 是一层薄的研发纪律加一个项目记忆闭环7 个入口:
| 用户输入 | 行为 |
|---|---|
| 明确行动诉求(修这个 bug、实现 X、重构 Y、审一下、记住这个… | **同轮直转**:报一句"按 `cs-xxx` 处理:{一句理由}",随即在当前回合按该 skill 的纪律继续执行,不要求用户重新调用或再次确认 |
| 咨询(该用哪个 / 流程怎么走 / 你建议怎么做) | 只推荐入口并说明理由,不启动执行 |
| 只说 cs、想了解体系、无具体诉求 | 输出体系速读 |
| 诉求含糊到无法判断行动类型,且选错会实质改变后续(会不会建档、会不会改代码) | 只问一个聚焦问题,不默认重流程 |
## 入口表
| 诉求 | 入口 |
|---|---|
@@ -22,10 +29,10 @@ CodeStable 是一层薄的研发纪律加一个项目记忆闭环7 个入口
| 沉淀经验、教训、"记住这个" | `cs-keep` |
| 仓库接入 / v1 升级 | `cs-onboard` |
项目记忆在 `.codestable/`attention.md每次必读、lessons/grep 检索的经验、work/(活动中的跨会话任务)。普通任务零产物,证据是 diff 与测试
一次只转一个入口;用户同时给出两个独立诉求时,问先做哪个。转入不扩大授权:目标 skill 的硬门槛、checkpoint 与写入规则照常生效
v1 的旧入口cs-feat-design、cs-issue-fix、cs-goal、cs-brainstorm、cs-docs、cs-domain、cs-req、cs-audit、cs-roadmap 等)已并入上表:设计与需求澄清是 cs-feat / cs-epic 的内置步骤,审计是 cs-code-review 的模式ADR 与文档由对应工作顺带完成或直接对话处理。
## 体系速读
## 回答方式
CodeStable 是一层薄研发纪律加一个项目记忆闭环。项目记忆在 `.codestable/`attention.md每次必读、lessons/grep 检索的经验、work/(活动中的跨会话任务)。普通任务零产物,证据是 diff 与测试。v1 的 24 个旧入口cs-feat-design、cs-goal、cs-audit、cs-note、cs-feedback、cs-roadmap 系等)已并入上表:设计与需求澄清是 cs-feat / cs-epic 的内置步骤,审计是 cs-code-review 的模式,沉淀统一走 cs-keep。
用户带着具体诉求来时,推荐一个入口并说明理由,让用户自己调用或同意后继续;诉求含糊时只问一个聚焦问题
导览与推荐本身不写任何文件
+1 -1
View File
@@ -13,7 +13,7 @@ SKILLS = ROOT / "plugins/codestable/skills"
LOCAL_SKILLS = ROOT / ".claude/skills"
THIN_SKILL_SAFETY_INVARIANTS = {
"cs": ("只解释和推荐", "不启动流程", "不写文件"),
"cs": ("同轮直转", "只推荐入口", "不写任何文件"),
"cs-code-review": ("只读", "独立 subagent reviewer", "blocking 未解决"),
"cs-epic": ("拆解方案必须经用户确认", "不代替用户做整体验收"),
"cs-feat": ("不得代替用户确认设计", "与声明相称的可核验证据"),