@devflow-tools/cli 0.18.27 → 0.18.28

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (42) hide show
  1. package/dist/plugin-files/.claude-plugin/plugin.json +1 -1
  2. package/dist/plugin-files/dist/command-registry.json +66 -50
  3. package/dist/plugin-files/dist/hooks/hook-daemon.js +26 -26
  4. package/dist/plugin-files/dist/hooks/post-tool-use-failure.js +2 -2
  5. package/dist/plugin-files/dist/hooks/post-tool-use.js +6 -6
  6. package/dist/plugin-files/dist/hooks/pre-compact.js +3 -3
  7. package/dist/plugin-files/dist/hooks/pre-tool-use.js +4 -4
  8. package/dist/plugin-files/dist/hooks/session-end.js +1 -1
  9. package/dist/plugin-files/dist/hooks/stop.js +9 -9
  10. package/dist/plugin-files/dist/hooks/user-prompt-submit.js +8 -8
  11. package/dist/plugin-files/dist/skills/devflow:animation/SKILL.md +1 -1
  12. package/dist/plugin-files/dist/skills/devflow:context/SKILL.md +9 -9
  13. package/dist/plugin-files/dist/skills/devflow:css/SKILL.md +1 -1
  14. package/dist/plugin-files/dist/skills/devflow:delivery/SKILL.md +25 -25
  15. package/dist/plugin-files/dist/skills/devflow:docker/SKILL.md +1 -1
  16. package/dist/plugin-files/dist/skills/devflow:doctor/SKILL.md +43 -0
  17. package/dist/plugin-files/dist/skills/devflow:electron/SKILL.md +1 -1
  18. package/dist/plugin-files/dist/skills/devflow:git/SKILL.md +1 -1
  19. package/dist/plugin-files/dist/skills/devflow:graph/SKILL.md +6 -6
  20. package/dist/plugin-files/dist/skills/devflow:graphql/SKILL.md +1 -1
  21. package/dist/plugin-files/dist/skills/devflow:knowledge/SKILL.md +6 -6
  22. package/dist/plugin-files/dist/skills/devflow:memory/SKILL.md +6 -6
  23. package/dist/plugin-files/dist/skills/devflow:nest/SKILL.md +2 -2
  24. package/dist/plugin-files/dist/skills/devflow:nextjs/SKILL.md +2 -2
  25. package/dist/plugin-files/dist/skills/devflow:performance/SKILL.md +1 -1
  26. package/dist/plugin-files/dist/skills/devflow:react/SKILL.md +5 -5
  27. package/dist/plugin-files/dist/skills/devflow:tailwind/SKILL.md +2 -2
  28. package/dist/plugin-files/dist/skills/devflow:taro/SKILL.md +1 -1
  29. package/dist/plugin-files/dist/skills/devflow:ui-layout/SKILL.md +1 -1
  30. package/dist/plugin-files/dist/skills/devflow:vue/SKILL.md +2 -2
  31. package/dist/plugin-files/dist/skills/devflow:workflow/SKILL.md +6 -6
  32. package/dist/plugin-files/hooks/session-start +10 -10
  33. package/dist/plugin-files/package.json +13 -13
  34. package/dist/plugin-files/skills/context/SKILL.md +7 -7
  35. package/dist/plugin-files/skills/delivery/SKILL.md +9 -9
  36. package/dist/plugin-files/skills/devflow/SKILL.md +1 -1
  37. package/dist/plugin-files/skills/doctor/SKILL.md +4 -1
  38. package/dist/plugin-files/skills/graph/SKILL.md +5 -5
  39. package/dist/plugin-files/skills/knowledge/SKILL.md +5 -5
  40. package/dist/plugin-files/skills/memory/SKILL.md +5 -5
  41. package/dist/plugin-files/skills/workflow/SKILL.md +5 -5
  42. package/package.json +28 -28
@@ -20,8 +20,8 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:context","required_mcp_tools":["ge
20
20
  ```
21
21
 
22
22
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
23
- - `mcp__devflow__get_project_context`
24
- - `mcp__devflow__search_symbol`
23
+ - `mcp__plugin_devflow_devflow__get_project_context`
24
+ - `mcp__plugin_devflow_devflow__search_symbol`
25
25
 
26
26
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
27
27
 
@@ -33,7 +33,7 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
33
33
  ## 工作流(必须按顺序执行)
34
34
 
35
35
  ### Step 1: 获取精确上下文
36
- 调用 `mcp__devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
36
+ 调用 `mcp__plugin_devflow_devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
37
37
  这一步确保你拿到的是项目真实状态,而非凭空猜测。
38
38
 
39
39
  ### Step 2: 分析
@@ -50,12 +50,12 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
50
50
 
51
51
  | 工具 | 何时调用 |
52
52
  |------|----------|
53
- | `mcp__devflow__get_project_context` | 每次任务第一步(必须) |
54
- | `mcp__devflow__get_project_context` | 按需调用 |
55
- | `mcp__devflow__search_symbol` | 按需调用 |
53
+ | `mcp__plugin_devflow_devflow__get_project_context` | 每次任务第一步(必须) |
54
+ | `mcp__plugin_devflow_devflow__get_project_context` | 按需调用 |
55
+ | `mcp__plugin_devflow_devflow__search_symbol` | 按需调用 |
56
56
 
57
57
  ## 规则
58
- - 永远从 `mcp__devflow__get_project_context` 开始——你不知道项目里有什么
58
+ - 永远从 `mcp__plugin_devflow_devflow__get_project_context` 开始——你不知道项目里有什么
59
59
  - 上下文返回后,用 Read 确认关键文件
60
60
  - 改完代码后跑类型检查
61
61
 
@@ -74,7 +74,7 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
74
74
  ## 执行
75
75
 
76
76
  1. 如果用户的查询包含中文,先将其中的中文关键词翻译为英文代码术语(空格分隔)
77
- 2. 调用 `mcp__devflow__get_project_context` MCP 工具:
77
+ 2. 调用 `mcp__plugin_devflow_devflow__get_project_context` MCP 工具:
78
78
 
79
79
  ```
80
80
  query = 用户的原始查询内容(包含中文)
@@ -82,7 +82,7 @@ expandedTerms = 你翻译的英文关键词(如 "code review audit module")
82
82
  mode = "fast" (首次调用;如需更深入结果,再改用 "deep")
83
83
  ```
84
84
 
85
- 也可以使用 `mcp__devflow__search_symbol` 按符号名搜索代码。
85
+ 也可以使用 `mcp__plugin_devflow_devflow__search_symbol` 按符号名搜索代码。
86
86
 
87
87
  示例:
88
88
  - 用户查询 "代码审核模块的流程" → expandedTerms = "code review audit module workflow pipeline process"
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:css","required_mcp_tools":["css_au
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__css_audit_selectors`
22
+ - `mcp__plugin_devflow_devflow__css_audit_selectors`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -34,22 +34,22 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:delivery","required_mcp_tools":["d
34
34
  ```
35
35
 
36
36
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
37
- - `mcp__devflow__delivery_status`
38
- - `mcp__devflow__delivery_create_case`
39
- - `mcp__devflow__delivery_answer`
40
- - `mcp__devflow__delivery_revise`
41
- - `mcp__devflow__delivery_compare_options`
42
- - `mcp__devflow__delivery_approve`
43
- - `mcp__devflow__delivery_reject`
44
- - `mcp__devflow__delivery_resume`
45
- - `mcp__devflow__delivery_recover_task`
46
- - `mcp__devflow__delivery_admit_worktree`
47
- - `mcp__devflow__delivery_refresh_activation`
48
- - `mcp__devflow__delivery_advance`
49
- - `mcp__devflow__delivery_retry_task`
50
- - `mcp__devflow__delivery_import_external_change`
51
- - `mcp__devflow__delivery_export`
52
- - `mcp__devflow__delivery_cancel`
37
+ - `mcp__plugin_devflow_devflow__delivery_status`
38
+ - `mcp__plugin_devflow_devflow__delivery_create_case`
39
+ - `mcp__plugin_devflow_devflow__delivery_answer`
40
+ - `mcp__plugin_devflow_devflow__delivery_revise`
41
+ - `mcp__plugin_devflow_devflow__delivery_compare_options`
42
+ - `mcp__plugin_devflow_devflow__delivery_approve`
43
+ - `mcp__plugin_devflow_devflow__delivery_reject`
44
+ - `mcp__plugin_devflow_devflow__delivery_resume`
45
+ - `mcp__plugin_devflow_devflow__delivery_recover_task`
46
+ - `mcp__plugin_devflow_devflow__delivery_admit_worktree`
47
+ - `mcp__plugin_devflow_devflow__delivery_refresh_activation`
48
+ - `mcp__plugin_devflow_devflow__delivery_advance`
49
+ - `mcp__plugin_devflow_devflow__delivery_retry_task`
50
+ - `mcp__plugin_devflow_devflow__delivery_import_external_change`
51
+ - `mcp__plugin_devflow_devflow__delivery_export`
52
+ - `mcp__plugin_devflow_devflow__delivery_cancel`
53
53
 
54
54
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
55
55
 
@@ -60,17 +60,17 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
60
60
 
61
61
  # Delivery Line
62
62
 
63
- 使用 `mcp__devflow__delivery_*` 工具推进一个持久化交付案件。案件状态和 `nextActions` 是唯一的流程事实来源。
63
+ 使用 `mcp__plugin_devflow_devflow__delivery_*` 工具推进一个持久化交付案件。案件状态和 `nextActions` 是唯一的流程事实来源。
64
64
 
65
65
  ## 开始案件
66
66
 
67
- 收到新的交付需求时,调用 `mcp__devflow__delivery_create_case`。使用当前项目绝对路径;默认 `baseRef=HEAD`、`targetRef=main`,除非用户明确指定其他值。为 actor、session、execution 和每次写操作生成稳定且可追踪的标识与幂等键。
67
+ 收到新的交付需求时,调用 `mcp__plugin_devflow_devflow__delivery_create_case`。使用当前项目绝对路径;默认 `baseRef=HEAD`、`targetRef=main`,除非用户明确指定其他值。为 actor、session、execution 和每次写操作生成稳定且可追踪的标识与幂等键。
68
68
  - 若上下文检索发现同名组件、同一路由或多个框架实现,将候选路径通过 `existingImplementations` 传入创建案件参数,让案件先进入阻塞澄清;不要自行选择复用或替换。
69
69
 
70
70
  ## 推进规则
71
71
 
72
- - 调用 `mcp__devflow__get_project_context` 后必须等待该 MCP 返回,再进行任何 Read/Grep/Glob/Bash/Write/Edit;不要并发发起上下文请求和直接工具调用。
73
- - 每次写操作后读取返回状态;状态不明确或发生 stale turn 时调用 `mcp__devflow__delivery_status`。
72
+ - 调用 `mcp__plugin_devflow_devflow__get_project_context` 后必须等待该 MCP 返回,再进行任何 Read/Grep/Glob/Bash/Write/Edit;不要并发发起上下文请求和直接工具调用。
73
+ - 每次写操作后读取返回状态;状态不明确或发生 stale turn 时调用 `mcp__plugin_devflow_devflow__delivery_status`。
74
74
  - 读取 `contextQuality.channelAvailability` 时区分三种结果:`available` 表示有可用证据,
75
75
  `empty_valid` 表示 producer 正常但项目没有匹配记录,可带降级说明继续;`unavailable`
76
76
  表示 producer 失败。Memory/Knowledge 默认是辅助通道,只有案件策略明确将其设为 required
@@ -94,11 +94,11 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
94
94
  - `delivery_advance` 仅允许在 `plan_approved` 阶段使用;其他阶段返回失败时必须按
95
95
  `nextActions` 恢复,不得把无推进当作成功。
96
96
  - 澄清与设计交互使用当前 turn;后续写操作携带案件当前 context receipt。
97
- - 只有用户明确同意时才能调用 `mcp__devflow__delivery_approve`;不得代替用户批准。
97
+ - 只有用户明确同意时才能调用 `mcp__plugin_devflow_devflow__delivery_approve`;不得代替用户批准。
98
98
  - 审批返回后不要结束会话:立即读取同一次返回的 `stage` 和 `nextActions`,按其允许的下一步继续推进;只有 `nextActions` 为空或用户明确暂停时才结束。
99
- - `mcp__devflow__delivery_advance` 用于推进已满足的状态,`mcp__devflow__delivery_resume` 只恢复已经具备运行条件的 gate 或 task。
100
- - 失败任务只通过 `mcp__devflow__delivery_retry_task` 重试;取消和导出分别使用 `mcp__devflow__delivery_cancel` 与 `mcp__devflow__delivery_export`。
101
- - Claude 在当前工作区产生的修改不是 Delivery 输出;只能先调用 `mcp__devflow__delivery_import_external_change`,由系统校验 base/candidate、允许文件、context receipt 和验证证据后再进入 integration。
99
+ - `mcp__plugin_devflow_devflow__delivery_advance` 用于推进已满足的状态,`mcp__plugin_devflow_devflow__delivery_resume` 只恢复已经具备运行条件的 gate 或 task。
100
+ - 失败任务只通过 `mcp__plugin_devflow_devflow__delivery_retry_task` 重试;取消和导出分别使用 `mcp__plugin_devflow_devflow__delivery_cancel` 与 `mcp__plugin_devflow_devflow__delivery_export`。
101
+ - Claude 在当前工作区产生的修改不是 Delivery 输出;只能先调用 `mcp__plugin_devflow_devflow__delivery_import_external_change`,由系统校验 base/candidate、允许文件、context receipt 和验证证据后再进入 integration。
102
102
 
103
103
  ## 可用操作
104
104
 
@@ -113,7 +113,7 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
113
113
  - MCP 返回 `ok: false` 时,该操作已经失败。向用户呈现 `error.code` 和 `error.message`,不得把 transport 成功误报为交付成功。
114
114
  - `error.retryable=false` 时不得自动重试;`error.recoveryAction` 是下一步建议,不是已完成的动作。
115
115
  - 遇到 `HELD`、`WORKER_UNAVAILABLE`、`REQUIRED_CONTEXT_MISSING` 或 producer/stale 警告时,先刷新 `delivery_status` 并保留阻塞状态;不得通过普通 Write/Edit/Bash/git commit 绕过 Delivery authority。
116
- - 遇到 `STALE_INTERACTION_TURN`、`MISSING_PREREQUISITE` 或 `INVALID_TRANSITION` 时,先调用 `mcp__devflow__delivery_status` 刷新案件状态;只有新的 `nextActions` 明确允许时才继续。
116
+ - 遇到 `STALE_INTERACTION_TURN`、`MISSING_PREREQUISITE` 或 `INVALID_TRANSITION` 时,先调用 `mcp__plugin_devflow_devflow__delivery_status` 刷新案件状态;只有新的 `nextActions` 明确允许时才继续。
117
117
  - Delivery 工具失败后不得为了绕过写入限制调用 `delivery_cancel`;只有用户明确要求取消时才可取消。取消或拒绝后的同一会话仍禁止直接 Write/Edit/Bash,必须新建 Delivery 案件或开启新会话。
118
118
  - 用户拒绝审批,或遇到 `UNAUTHORIZED_ROLE`、权限失败、`PROJECT_ROOT_MISMATCH` 时不得自动重试;保留案件当前状态并请求用户处理。
119
119
  - 不通过更换 actor、project root、turn、context receipt 或幂等键来规避失败。
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:docker","required_mcp_tools":["doc
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__docker_dockerize_project`
22
+ - `mcp__plugin_devflow_devflow__docker_dockerize_project`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -0,0 +1,43 @@
1
+ ---
2
+ name: devflow:doctor
3
+ description: 项目健康检查 — 索引状态、引擎就绪、插件列表
4
+ required_mcp_tools:
5
+ - devflow_runtime_status
6
+ ---
7
+
8
+ <HARD-GATE>
9
+ 收到 devflow:doctor 时,必须先调用以下 MCP 工具之一:
10
+ devflow_runtime_status
11
+ </HARD-GATE>
12
+
13
+ ## MANDATORY FIRST STEP - Skill Registration
14
+
15
+ > **IMPORTANT: Before ANY other action, execute the registration command below. This is NON-NEGOTIABLE.**
16
+
17
+ ```bash
18
+ mkdir -p ~/.devflow && echo '{"name":"devflow:doctor","required_mcp_tools":["devflow_runtime_status"],"description":"devflow:doctor skill","registeredAt":'"$(date +%s)000"'}' > ~/.devflow/current-skill.json
19
+ ```
20
+
21
+ After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
+ - `mcp__plugin_devflow_devflow__devflow_runtime_status`
23
+
24
+ **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
+
26
+ When the skill execution is complete, clean up:
27
+ ```bash
28
+ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
29
+ ```
30
+
31
+ # 项目健康检查
32
+
33
+ 快速了解 DevFlow 在当前项目中的运行状态。任务开始前用一次即可。
34
+
35
+ 调用 `mcp__plugin_devflow_devflow__devflow_runtime_status` 检查当前 Claude session、task turn 和语义上下文是否正确绑定。该工具是只读控制面诊断,不依赖已有 context receipt。
36
+
37
+ ## 检查项
38
+
39
+ - 文件索引是否最新
40
+ - Context/Graph/Memory/Knowledge/Workflow 引擎是否就绪
41
+ - 已安装哪些领域插件
42
+ - 是否有已知配置问题
43
+ **MCP Gateway 策略**: 调用前会校验 Session Token 和调用合法性。在支持 PreToolUse Hook 的 Claude Code 等宿主中,无有效收据的原生工具调用会被拦截;不提供生命周期 Hook 的宿主不会声称已执行原生拦截,而是依赖 canonical MCP receipts 和显式 capability degradation。收据有效期 30 秒,过期后需重新通过 MCP 工具获取。
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:electron","required_mcp_tools":["e
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__electron_diagnose_bug`
22
+ - `mcp__plugin_devflow_devflow__electron_diagnose_bug`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:git","required_mcp_tools":["git_an
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__git_analyze_repo`
22
+ - `mcp__plugin_devflow_devflow__git_analyze_repo`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:graph","required_mcp_tools":["get_
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__get_dependency_graph`
22
+ - `mcp__plugin_devflow_devflow__get_dependency_graph`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -31,7 +31,7 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
31
31
  ## 工作流(必须按顺序执行)
32
32
 
33
33
  ### Step 1: 获取精确上下文
34
- 调用 `mcp__devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
34
+ 调用 `mcp__plugin_devflow_devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
35
35
  这一步确保你拿到的是项目真实状态,而非凭空猜测。
36
36
 
37
37
  ### Step 2: 分析
@@ -48,11 +48,11 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
48
48
 
49
49
  | 工具 | 何时调用 |
50
50
  |------|----------|
51
- | `mcp__devflow__get_project_context` | 每次任务第一步(必须) |
52
- | `mcp__devflow__get_dependency_graph` | 按需调用 |
51
+ | `mcp__plugin_devflow_devflow__get_project_context` | 每次任务第一步(必须) |
52
+ | `mcp__plugin_devflow_devflow__get_dependency_graph` | 按需调用 |
53
53
 
54
54
  ## 规则
55
- - 永远从 `mcp__devflow__get_project_context` 开始——你不知道项目里有什么
55
+ - 永远从 `mcp__plugin_devflow_devflow__get_project_context` 开始——你不知道项目里有什么
56
56
  - 上下文返回后,用 Read 确认关键文件
57
57
  - 改完代码后跑类型检查
58
58
 
@@ -70,7 +70,7 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
70
70
 
71
71
  ## 执行
72
72
 
73
- 调用 `mcp__devflow__get_dependency_graph` MCP 工具:
73
+ 调用 `mcp__plugin_devflow_devflow__get_dependency_graph` MCP 工具:
74
74
 
75
75
  ```
76
76
  target = "all" (默认,拉取完整图谱)
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:graphql","required_mcp_tools":["gr
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__graphql_diagnose_bug`
22
+ - `mcp__plugin_devflow_devflow__graphql_diagnose_bug`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:knowledge","required_mcp_tools":["
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__get_knowledge`
22
+ - `mcp__plugin_devflow_devflow__get_knowledge`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -31,7 +31,7 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
31
31
  ## 工作流(必须按顺序执行)
32
32
 
33
33
  ### Step 1: 获取精确上下文
34
- 调用 `mcp__devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
34
+ 调用 `mcp__plugin_devflow_devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
35
35
  这一步确保你拿到的是项目真实状态,而非凭空猜测。
36
36
 
37
37
  ### Step 2: 分析
@@ -48,11 +48,11 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
48
48
 
49
49
  | 工具 | 何时调用 |
50
50
  |------|----------|
51
- | `mcp__devflow__get_project_context` | 每次任务第一步(必须) |
52
- | `mcp__devflow__get_knowledge` | 按需调用 |
51
+ | `mcp__plugin_devflow_devflow__get_project_context` | 每次任务第一步(必须) |
52
+ | `mcp__plugin_devflow_devflow__get_knowledge` | 按需调用 |
53
53
 
54
54
  ## 规则
55
- - 永远从 `mcp__devflow__get_project_context` 开始——你不知道项目里有什么
55
+ - 永远从 `mcp__plugin_devflow_devflow__get_project_context` 开始——你不知道项目里有什么
56
56
  - 上下文返回后,用 Read 确认关键文件
57
57
  - 改完代码后跑类型检查
58
58
 
@@ -70,7 +70,7 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
70
70
 
71
71
  ## 执行
72
72
 
73
- 调用 `mcp__devflow__get_knowledge` MCP 工具:
73
+ 调用 `mcp__plugin_devflow_devflow__get_knowledge` MCP 工具:
74
74
 
75
75
  ```
76
76
  query = 用户的查询内容
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:memory","required_mcp_tools":["get
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__get_memory`
22
+ - `mcp__plugin_devflow_devflow__get_memory`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -31,7 +31,7 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
31
31
  ## 工作流(必须按顺序执行)
32
32
 
33
33
  ### Step 1: 获取精确上下文
34
- 调用 `mcp__devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
34
+ 调用 `mcp__plugin_devflow_devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
35
35
  这一步确保你拿到的是项目真实状态,而非凭空猜测。
36
36
 
37
37
  ### Step 2: 分析
@@ -48,11 +48,11 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
48
48
 
49
49
  | 工具 | 何时调用 |
50
50
  |------|----------|
51
- | `mcp__devflow__get_project_context` | 每次任务第一步(必须) |
52
- | `mcp__devflow__get_memory` | 按需调用 |
51
+ | `mcp__plugin_devflow_devflow__get_project_context` | 每次任务第一步(必须) |
52
+ | `mcp__plugin_devflow_devflow__get_memory` | 按需调用 |
53
53
 
54
54
  ## 规则
55
- - 永远从 `mcp__devflow__get_project_context` 开始——你不知道项目里有什么
55
+ - 永远从 `mcp__plugin_devflow_devflow__get_project_context` 开始——你不知道项目里有什么
56
56
  - 上下文返回后,用 Read 确认关键文件
57
57
  - 改完代码后跑类型检查
58
58
 
@@ -70,7 +70,7 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
70
70
 
71
71
  ## 执行
72
72
 
73
- 调用 `mcp__devflow__get_memory` MCP 工具(无需参数):
73
+ 调用 `mcp__plugin_devflow_devflow__get_memory` MCP 工具(无需参数):
74
74
 
75
75
  ```
76
76
  (无参数 — 返回技术栈、约定、偏好、历史决策)
@@ -20,8 +20,8 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:nest","required_mcp_tools":["nest_
20
20
  ```
21
21
 
22
22
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
23
- - `mcp__devflow__nest_diagnose_bug`
24
- - `mcp__devflow__nest_new_module`
23
+ - `mcp__plugin_devflow_devflow__nest_diagnose_bug`
24
+ - `mcp__plugin_devflow_devflow__nest_new_module`
25
25
 
26
26
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
27
27
 
@@ -20,8 +20,8 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:nextjs","required_mcp_tools":["nex
20
20
  ```
21
21
 
22
22
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
23
- - `mcp__devflow__nextjs_diagnose_bug`
24
- - `mcp__devflow__nextjs_new_page`
23
+ - `mcp__plugin_devflow_devflow__nextjs_diagnose_bug`
24
+ - `mcp__plugin_devflow_devflow__nextjs_new_page`
25
25
 
26
26
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
27
27
 
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:performance","required_mcp_tools":
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__performance_audit_performance`
22
+ - `mcp__plugin_devflow_devflow__performance_audit_performance`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -23,11 +23,11 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:react","required_mcp_tools":["reac
23
23
  ```
24
24
 
25
25
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
26
- - `mcp__devflow__react_review_hooks`
27
- - `mcp__devflow__react_audit_performance`
28
- - `mcp__devflow__react_refactor_component`
29
- - `mcp__devflow__react_diagnose_bug`
30
- - `mcp__devflow__react_new_component`
26
+ - `mcp__plugin_devflow_devflow__react_review_hooks`
27
+ - `mcp__plugin_devflow_devflow__react_audit_performance`
28
+ - `mcp__plugin_devflow_devflow__react_refactor_component`
29
+ - `mcp__plugin_devflow_devflow__react_diagnose_bug`
30
+ - `mcp__plugin_devflow_devflow__react_new_component`
31
31
 
32
32
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
33
33
 
@@ -20,8 +20,8 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:tailwind","required_mcp_tools":["t
20
20
  ```
21
21
 
22
22
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
23
- - `mcp__devflow__tailwind_new_component`
24
- - `mcp__devflow__tailwind_optimize_classes`
23
+ - `mcp__plugin_devflow_devflow__tailwind_new_component`
24
+ - `mcp__plugin_devflow_devflow__tailwind_optimize_classes`
25
25
 
26
26
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
27
27
 
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:taro","required_mcp_tools":["taro_
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__taro_new_page`
22
+ - `mcp__plugin_devflow_devflow__taro_new_page`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:ui-layout","required_mcp_tools":["
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__ui_layout_generate`
22
+ - `mcp__plugin_devflow_devflow__ui_layout_generate`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -20,8 +20,8 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:vue","required_mcp_tools":["vue_di
20
20
  ```
21
21
 
22
22
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
23
- - `mcp__devflow__vue_diagnose_bug`
24
- - `mcp__devflow__vue_debug_reactivity`
23
+ - `mcp__plugin_devflow_devflow__vue_diagnose_bug`
24
+ - `mcp__plugin_devflow_devflow__vue_debug_reactivity`
25
25
 
26
26
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
27
27
 
@@ -19,7 +19,7 @@ mkdir -p ~/.devflow && echo '{"name":"devflow:workflow","required_mcp_tools":["r
19
19
  ```
20
20
 
21
21
  After registration, you **MUST** use ONLY these MCP tools for all work in this skill:
22
- - `mcp__devflow__run_workflow`
22
+ - `mcp__plugin_devflow_devflow__run_workflow`
23
23
 
24
24
  **Never use direct tools** (Read, Write, Edit, Bash, Grep, etc.) for tasks that can be handled by the MCP tools above.
25
25
 
@@ -31,7 +31,7 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
31
31
  ## 工作流(必须按顺序执行)
32
32
 
33
33
  ### Step 1: 获取精确上下文
34
- 调用 `mcp__devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
34
+ 调用 `mcp__plugin_devflow_devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
35
35
  这一步确保你拿到的是项目真实状态,而非凭空猜测。
36
36
 
37
37
  ### Step 2: 分析
@@ -48,11 +48,11 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
48
48
 
49
49
  | 工具 | 何时调用 |
50
50
  |------|----------|
51
- | `mcp__devflow__get_project_context` | 每次任务第一步(必须) |
52
- | `mcp__devflow__run_workflow` | 按需调用 |
51
+ | `mcp__plugin_devflow_devflow__get_project_context` | 每次任务第一步(必须) |
52
+ | `mcp__plugin_devflow_devflow__run_workflow` | 按需调用 |
53
53
 
54
54
  ## 规则
55
- - 永远从 `mcp__devflow__get_project_context` 开始——你不知道项目里有什么
55
+ - 永远从 `mcp__plugin_devflow_devflow__get_project_context` 开始——你不知道项目里有什么
56
56
  - 上下文返回后,用 Read 确认关键文件
57
57
  - 改完代码后跑类型检查
58
58
 
@@ -70,7 +70,7 @@ rm -f ~/.devflow/current-skill.json ~/.devflow/current-execution-id
70
70
 
71
71
  ## 执行
72
72
 
73
- 调用 `mcp__devflow__run_workflow` MCP 工具:
73
+ 调用 `mcp__plugin_devflow_devflow__run_workflow` MCP 工具:
74
74
 
75
75
  ```
76
76
  type = "bugfix" | "feature" | "review"
@@ -81,33 +81,33 @@ MCP_PROTOCOL="
81
81
  当用户使用 /devflow: 开头的技能时(如 /devflow:react、/devflow:vue、/devflow:context 等),你必须按以下流程操作:
82
82
 
83
83
  ### Step 1: 获取上下文(必须先执行)
84
- 调用 mcp__devflow__get_project_context,参数格式:{\"query\": \"用户的问题描述\", \"projectRoot\": \"\${CLAUDE_PROJECT_DIR}\"}
84
+ 调用 mcp__plugin_devflow_devflow__get_project_context,参数格式:{\"query\": \"用户的问题描述\", \"projectRoot\": \"\${CLAUDE_PROJECT_DIR}\"}
85
85
  这个工具返回项目中的相关文件、符号、风险提示和下一步建议。
86
86
  不要跳过这一步。不要用 Explore、Agent、Grep 或 Bash 代替。
87
87
 
88
88
  ### Step 2: 按需调用领域工具
89
89
  根据已加载 Skill 和 command registry 调用对应的 MCP 工具。以下仅为高频示例,完整能力以 /devflow 列表和 command registry 为准:
90
- - React 性能问题 → mcp__devflow__react_audit_performance(参数:{\"projectRoot\": \"\${CLAUDE_PROJECT_DIR}\"})
91
- - React Bug → mcp__devflow__react_diagnose_bug
92
- - React Hooks 审查 → mcp__devflow__react_review_hooks
93
- - Vue 问题 → mcp__devflow__vue_diagnose_bug
94
- - NestJS 问题 → mcp__devflow__nest_diagnose_bug
95
- - 代码搜索 → mcp__devflow__search_symbol
96
- - 知识查询 → mcp__devflow__get_knowledge
90
+ - React 性能问题 → mcp__plugin_devflow_devflow__react_audit_performance(参数:{\"projectRoot\": \"\${CLAUDE_PROJECT_DIR}\"})
91
+ - React Bug → mcp__plugin_devflow_devflow__react_diagnose_bug
92
+ - React Hooks 审查 → mcp__plugin_devflow_devflow__react_review_hooks
93
+ - Vue 问题 → mcp__plugin_devflow_devflow__vue_diagnose_bug
94
+ - NestJS 问题 → mcp__plugin_devflow_devflow__nest_diagnose_bug
95
+ - 代码搜索 → mcp__plugin_devflow_devflow__search_symbol
96
+ - 知识查询 → mcp__plugin_devflow_devflow__get_knowledge
97
97
 
98
98
  ### Step 3: 基于工具返回结果进行分析
99
99
  用 Read 确认工具返回的关键文件,然后给出分析结论。
100
100
 
101
101
  ### 铁律
102
102
  - 永远不要用 Explore、Agent、Grep 或 Glob 替代 MCP 工具的第一步上下文获取
103
- - Every mcp__devflow__ tool MUST pass projectRoot = \${CLAUDE_PROJECT_DIR}; missing projectRoot is rejected; no cwd fallback.
103
+ - Every mcp__plugin_devflow_devflow__ tool MUST pass projectRoot = \${CLAUDE_PROJECT_DIR}; missing projectRoot is rejected; no cwd fallback.
104
104
  - 如果 MCP 工具返回错误,重试一次;再失败则回退到直接读文件"
105
105
 
106
106
  COMPACT_PROTOCOL="
107
107
 
108
108
  ## DevFlow compact protocol
109
109
  - Project: ${PROJECT_NAME} (${PROJECT_ROOT})
110
- - For /devflow:* tasks, call mcp__devflow__get_project_context before broad native search, then use the applicable domain MCP tool.
110
+ - For /devflow:* tasks, call mcp__plugin_devflow_devflow__get_project_context before broad native search, then use the applicable domain MCP tool.
111
111
  - Explicit memory is successful only after a DevFlow canonical receipt. Queue acceptance or host-local memory is not a successful save.
112
112
  - Static findings are not runtime measurements. Do not claim observed bottlenecks, severity, or quantified benefit without evidence.
113
113
  - Workflow success requires canonical host-action and verification evidence. Never convert degraded, cancelled, failed, or unknown state into success.
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@devflow-tools/claude-code-plugin",
3
- "version": "0.18.27",
3
+ "version": "0.18.28",
4
4
  "description": "Claude Code hooks and skills for the DevFlow development intelligence runtime.",
5
5
  "type": "module",
6
6
  "scripts": {
@@ -12,20 +12,20 @@
12
12
  "verify:runtime-dependencies": "node --import tsx scripts/verify-runtime-dependencies.ts"
13
13
  },
14
14
  "dependencies": {
15
- "@devflow-tools/benchmark": "0.18.27",
16
- "@devflow-tools/context-engine": "0.18.27",
17
- "@devflow-tools/database": "0.18.27",
18
- "@devflow-tools/mcp-server": "0.18.27",
19
- "@devflow-tools/memory-engine": "0.18.27",
20
- "@devflow-tools/sdk": "0.18.27",
21
- "@devflow-tools/semantic-engine": "0.18.27",
22
- "@devflow-tools/server": "0.18.27",
23
- "@devflow-tools/task-runtime": "0.18.27",
24
- "@devflow-tools/telemetry": "0.18.27",
25
- "@devflow-tools/workflow-engine": "0.18.27"
15
+ "@devflow-tools/benchmark": "0.18.28",
16
+ "@devflow-tools/context-engine": "0.18.28",
17
+ "@devflow-tools/database": "0.18.28",
18
+ "@devflow-tools/mcp-server": "0.18.28",
19
+ "@devflow-tools/memory-engine": "0.18.28",
20
+ "@devflow-tools/sdk": "0.18.28",
21
+ "@devflow-tools/semantic-engine": "0.18.28",
22
+ "@devflow-tools/server": "0.18.28",
23
+ "@devflow-tools/task-runtime": "0.18.28",
24
+ "@devflow-tools/telemetry": "0.18.28",
25
+ "@devflow-tools/workflow-engine": "0.18.28"
26
26
  },
27
27
  "devDependencies": {
28
- "@devflow-tools/devflow": "0.18.27"
28
+ "@devflow-tools/devflow": "0.18.28"
29
29
  },
30
30
  "allowScripts": {
31
31
  "better-sqlite3@12.11.1": true,
@@ -6,7 +6,7 @@ description: 检索项目上下文 — 自然语言查询相关代码、符号
6
6
  ## 工作流(必须按顺序执行)
7
7
 
8
8
  ### Step 1: 获取精确上下文
9
- 调用 `mcp__devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
9
+ 调用 `mcp__plugin_devflow_devflow__get_project_context` 获取与用户问题相关的代码、记忆和知识库信息。
10
10
  这一步确保你拿到的是项目真实状态,而非凭空猜测。
11
11
 
12
12
  ### Step 2: 分析
@@ -23,12 +23,12 @@ description: 检索项目上下文 — 自然语言查询相关代码、符号
23
23
 
24
24
  | 工具 | 何时调用 |
25
25
  |------|----------|
26
- | `mcp__devflow__get_project_context` | 每次任务第一步(必须) |
27
- | `mcp__devflow__get_project_context` | 按需调用 |
28
- | `mcp__devflow__search_symbol` | 按需调用 |
26
+ | `mcp__plugin_devflow_devflow__get_project_context` | 每次任务第一步(必须) |
27
+ | `mcp__plugin_devflow_devflow__get_project_context` | 按需调用 |
28
+ | `mcp__plugin_devflow_devflow__search_symbol` | 按需调用 |
29
29
 
30
30
  ## 规则
31
- - 永远从 `mcp__devflow__get_project_context` 开始——你不知道项目里有什么
31
+ - 永远从 `mcp__plugin_devflow_devflow__get_project_context` 开始——你不知道项目里有什么
32
32
  - 上下文返回后,用 Read 确认关键文件
33
33
  - 改完代码后跑类型检查
34
34
 
@@ -47,7 +47,7 @@ description: 检索项目上下文 — 自然语言查询相关代码、符号
47
47
  ## 执行
48
48
 
49
49
  1. 如果用户的查询包含中文,先将其中的中文关键词翻译为英文代码术语(空格分隔)
50
- 2. 调用 `mcp__devflow__get_project_context` MCP 工具:
50
+ 2. 调用 `mcp__plugin_devflow_devflow__get_project_context` MCP 工具:
51
51
 
52
52
  ```
53
53
  query = 用户的原始查询内容(包含中文)
@@ -55,7 +55,7 @@ expandedTerms = 你翻译的英文关键词(如 "code review audit module")
55
55
  mode = "fast" (首次调用;如需更深入结果,再改用 "deep")
56
56
  ```
57
57
 
58
- 也可以使用 `mcp__devflow__search_symbol` 按符号名搜索代码。
58
+ 也可以使用 `mcp__plugin_devflow_devflow__search_symbol` 按符号名搜索代码。
59
59
 
60
60
  示例:
61
61
  - 用户查询 "代码审核模块的流程" → expandedTerms = "code review audit module workflow pipeline process"