ln-400-story-executor by levnikolaevich/claude-code-skills
npx skills add https://github.com/levnikolaevich/claude-code-skills --skill ln-400-story-executorPaths: File paths (
shared/,references/,../ln-*) are relative to skills repo root. If not found at CWD, locate this SKILL.md directory and go up one level for repo root. Ifshared/is missing, fetch files via WebFetch fromhttps://raw.githubusercontent.com/levnikolaevich/claude-code-skills/master/skills/{path}.
Type: L2 Coordinator Category: 4XX Execution
通过按优先级顺序循环处理故事中的任务,端到端地执行一个故事。当所有任务完成时(质量门决定完成),将故事状态设置为 待审核。
| 输入 | 必需 | 来源 |
|---|
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
| 描述 |
|---|
storyId | 是 | args, git branch, kanban, user | 要处理的故事 |
解析方式: 故事解析链。状态过滤器: 待办, 进行中
必须阅读: 加载 shared/references/tools_config_guide.md、shared/references/storage_mode_detection.md 和 shared/references/input_resolution_pattern.md
提取:task_provider = 任务管理 → 提供者 (linear | file)。
必须阅读: 加载 shared/references/git_worktree_fallback.md — 使用"故事执行"行。
feature/{identifier}-{story-title-slug}(小写,空格转短横线,无特殊字符)git branch --show-current
feature/* 分支上 → 使用当前工作树,跳转到阶段 2.hex-skills/worktrees/story-{identifier})获取故事元数据和所有子任务元数据(仅限 ID/标题/状态/标签):
list_issues(parentId=Story.id)Glob("docs/tasks/epics/*/stories/{story-slug}/tasks/*.md") + 解析 **Status:**汇总计数(例如,"2 个待审核,1 个待返工,3 个待办")。不进行分析 — 立即继续。
在委托待办任务之前,根据当前代码库验证其计划:
跳过上下文审核的情况: 待审核任务,待返工任务,当实现任务刚完成时的测试任务,创建时间 <24 小时的任务。
优先级顺序: 待审核 > 待返工 > 待办(在每个状态内优先处理基础任务)。
基于分组的待办任务分派:
**Parallel Group:** {N}。没有此字段的任务 = 每个任务自成一组(顺序执行,向后兼容)。执行 → 审核 → 下一个。 永不跳过审核。审核始终是顺序进行的(ln-402 内联执行)。
| 条件 | 操作 |
|---|---|
| 所有任务状态 = 完成 | 停止 — 进入阶段 5(完成) |
| 同一任务连续 3 次以上处于待返工状态 | 停止 — 上报:"检测到任务返工循环,需要输入" |
| 没有处于可处理状态(待审核 / 待返工 / 待办)的任务 | 停止 — 进入阶段 5 |
当所有任务完成时:
update_issue(id, state: "To Review");文件:编辑 **Status:** 行)关键: 执行器(ln-401/ln-403/ln-404)使用 Agent 工具进行上下文隔离。审核器(ln-402)在主流程中通过 Skill 工具内联运行。
| 状态 | 工作器 | 备注 |
|---|---|---|
| 待审核 | ln-402-task-reviewer | 内联(Skill 工具)。 按 ID 加载任务,在主流程中审核。无子代理。 |
| 待返工 | ln-403-task-rework | 然后立即对同一任务调用 ln-402 |
| 待办(测试) | ln-404-test-executor | 然后立即对同一任务调用 ln-402 |
| 待办(实现) | ln-401-task-executor | 然后立即对同一任务调用 ln-402 |
| 提示词模板: |
执行器(ln-401/ln-403/ln-404)— Agent 工具(隔离上下文):
Agent(description: "[Action] task {ID}",
prompt: "Execute task {ID}.
Step 1: Invoke worker:
Skill(skill: \"{skill-name}\")
CONTEXT:
Task ID: {ID}",
subagent_type: "general-purpose")
审核器(ln-402)— Skill 工具(主流程):
Skill(skill: "ln-402-task-reviewer", args: "{task-ID}")
在每个任务之前,添加以下两个步骤:
Execute [Task-ID]: [Title] — 开始时标记为进行中Review [Task-ID]: [Title] — 执行器完成后标记为进行中,ln-402 完成后标记为已完成feature/{story-id}-{slug} 分支的隔离工作树中进行。切勿修改主工作树。**Parallel Group:** 字段,则回退到完全顺序执行。mypy/ruff/pytest 而不是调用技能零容忍: 如果直接运行命令而不是调用技能,请停止并纠正。
当在计划模式(代理无法执行)下调用时,生成执行计划:
计划输出格式:
## Execution Plan for Story {STORY-ID}: {Title}
| # | Task ID | Title | Status | Group | Executor | Reviewer |
|---|---------|-------|--------|-------|----------|----------|
| 1 | {ID} | {Title} | {Status} | {N} | ln-40X | ln-402 |
### Sequence
1. [Execute] {Task-1} via ln-401-task-executor
2. [Review] {Task-1} via ln-402-task-reviewer
...
必须阅读: 加载 shared/references/meta_analysis_protocol.md
技能类型:execution-orchestrator。在所有阶段完成后运行。使用 execution-orchestrator 格式输出到聊天。
shared/references/tools_config_guide.mdshared/references/storage_mode_detection.mdshared/references/orchestrator_pattern.mdshared/references/task_delegation_pattern.mdshared/references/auto_discovery_pattern.mdshared/references/plan_mode_pattern.mdshared/references/git_worktree_fallback.md../ln-401-task-executor/SKILL.md, ../ln-403-task-rework/SKILL.md, ../ln-404-test-executor/SKILL.md../ln-402-task-reviewer/SKILL.mdCLAUDE.md, docs/tasks/kanban_board.mdVersion: 4.0.0 Last Updated: 2026-01-29
每周安装次数
152
代码仓库
GitHub 星标数
245
首次出现
Jan 24, 2026
安全审计
安装于
claude-code138
gemini-cli136
codex136
cursor136
opencode136
github-copilot131
Paths: File paths (
shared/,references/,../ln-*) are relative to skills repo root. If not found at CWD, locate this SKILL.md directory and go up one level for repo root. Ifshared/is missing, fetch files via WebFetch fromhttps://raw.githubusercontent.com/levnikolaevich/claude-code-skills/master/skills/{path}.
Type: L2 Coordinator Category: 4XX Execution
Executes a Story end-to-end by looping through its tasks in priority order. Sets Story to To Review when all tasks Done (quality gate decides Done).
| Input | Required | Source | Description |
|---|---|---|---|
storyId | Yes | args, git branch, kanban, user | Story to process |
Resolution: Story Resolution Chain. Status filter: Todo, In Progress
MANDATORY READ: Load shared/references/tools_config_guide.md, shared/references/storage_mode_detection.md, and shared/references/input_resolution_pattern.md
Extract: task_provider = Task Management → Provider (linear | file).
MANDATORY READ: Load shared/references/git_worktree_fallback.md — use "Story execution" row.
feature/{identifier}-{story-title-slug} (lowercase, spaces→dashes, no special chars)git branch --show-current
feature/* → use current worktree, skip to Phase 2.hex-skills/worktrees/story-{identifier})Fetch Story metadata and all child task metadata (ID/title/status/labels only):
list_issues(parentId=Story.id)Glob("docs/tasks/epics/*/stories/{story-slug}/tasks/*.md") + parse **Status:**Summarize counts (e.g., "2 To Review, 1 To Rework, 3 Todo"). NO analysis — proceed immediately.
Before delegating a Todo task, verify its plan against current codebase:
Skip Context Review for: To Review tasks, To Rework tasks, test tasks when impl freshly Done, tasks created <24h ago.
Priority order: To Review > To Rework > Todo (foundation-first within each status).
Group-based dispatch for Todo tasks:
**Parallel Group:** {N} from each Todo task. Tasks without this field = each gets its own group (sequential, backward compatible).Execute → Review → Next. Never skip review. Reviews are always sequential (ln-402 inline).
| Condition | Action |
|---|---|
| All tasks status = Done | STOP — proceed to Phase 5 (Completion) |
| Same task in To Rework 3+ consecutive times | STOP — ESCALATE: "Task rework loop detected, need input" |
| No tasks in processable status (To Review / To Rework / Todo) | STOP — proceed to Phase 5 |
When all tasks Done:
update_issue(id, state: "To Review"); File: Edit the **Status:** line)CRITICAL: Executors (ln-401/ln-403/ln-404) use Agent tool for context isolation. Reviewer (ln-402) runs inline via Skill tool in main flow.
| Status | Worker | Notes |
|---|---|---|
| To Review | ln-402-task-reviewer | Inline (Skill tool). Load task by ID, review in main flow. No subagent. |
| To Rework | ln-403-task-rework | Then immediate ln-402 on same task |
| Todo (tests) | ln-404-test-executor | Then immediate ln-402 on same task |
| Todo (impl) | ln-401-task-executor | Then immediate ln-402 on same task |
| Prompt templates: |
Executors (ln-401/ln-403/ln-404) — Agent tool (isolated context):
Agent(description: "[Action] task {ID}",
prompt: "Execute task {ID}.
Step 1: Invoke worker:
Skill(skill: \"{skill-name}\")
CONTEXT:
Task ID: {ID}",
subagent_type: "general-purpose")
Reviewer (ln-402) — Skill tool (main flow):
Skill(skill: "ln-402-task-reviewer", args: "{task-ID}")
Before each task, add BOTH steps:
Execute [Task-ID]: [Title] — mark in_progress when startingReview [Task-ID]: [Title] — mark in_progress after executor, completed after ln-402feature/{story-id}-{slug}. Never modify main worktree**Parallel Group:** missing on any task, fall back to fully sequential execution.mypy/ruff/pytest directly instead of skill invocationZERO TOLERANCE: If running commands directly instead of invoking skills, STOP and correct.
When invoked in Plan Mode (agent cannot execute), generate execution plan instead:
Plan Output Format:
## Execution Plan for Story {STORY-ID}: {Title}
| # | Task ID | Title | Status | Group | Executor | Reviewer |
|---|---------|-------|--------|-------|----------|----------|
| 1 | {ID} | {Title} | {Status} | {N} | ln-40X | ln-402 |
### Sequence
1. [Execute] {Task-1} via ln-401-task-executor
2. [Review] {Task-1} via ln-402-task-reviewer
...
MANDATORY READ: Load shared/references/meta_analysis_protocol.md
Skill type: execution-orchestrator. Run after all phases complete. Output to chat using the execution-orchestrator format.
shared/references/tools_config_guide.mdshared/references/storage_mode_detection.mdshared/references/orchestrator_pattern.mdshared/references/task_delegation_pattern.mdshared/references/auto_discovery_pattern.mdshared/references/plan_mode_pattern.mdshared/references/git_worktree_fallback.mdVersion: 4.0.0 Last Updated: 2026-01-29
Weekly Installs
152
Repository
GitHub Stars
245
First Seen
Jan 24, 2026
Security Audits
Gen Agent Trust HubPassSocketPassSnykWarn
Installed on
claude-code138
gemini-cli136
codex136
cursor136
opencode136
github-copilot131
通过 LiteLLM 代理让 Claude Code 对接 GitHub Copilot 运行 | 高级变通方案指南
36,300 周安装
../ln-401-task-executor/SKILL.md, ../ln-403-task-rework/SKILL.md, ../ln-404-test-executor/SKILL.md../ln-402-task-reviewer/SKILL.mdCLAUDE.md, docs/tasks/kanban_board.md