Token导航 LogoToken导航

Agent Skill · 开发

workflow-patterns(工作流程模式)

workflow-patterns 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

workflow-patterns 是开发类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为执行命令;许可证信息还不明确,第三方扫描已通过,按安装量在开发类的 21,746 个 Skill 里排第 20,285。

总安装 233周安装 10GitHub Stars 2下载 82核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:workflow-patterns(工作流程模式)
来源仓库:https://github.com/masanao-ohba/claude-manifests
仓库路径:skills/workflow-patterns
安装命令:
npx skills add https://github.com/masanao-ohba/claude-manifests --skill workflow-patterns
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/masanao-ohba/claude-manifests --skill workflow-patterns

这个 Skill 是做什么的

开发类 Skill 面向写代码、改代码和排查问题,一般会读写仓库文件,部分还会执行构建或测试命令。

workflow-patterns 用于处理 GitHub 仓库、Issue 和 Pull Request 协作信息,适合围绕代码变更进行整理。适用于仓库状态跟踪、代码变更分析和协作事项管理场景。通过 GitHub API 集成实现仓库信息和协作数据的获取。

安装前需确认 GitHub 访问权限和维护状态,注意可能触发网络请求。建议结合原始 README 核验具体用法,确保符合实际使用边界。

来源仓库内路径:skills/workflow-patterns

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/masanao-ohba/claude-manifests --skill workflow…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要执行命令,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

workflow-patterns 怎么安装?

在终端执行 npx skills add https://github.com/masanao-ohba/claude-manifests --skill workflow-patterns。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

workflow-patterns 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

workflow-patterns 需要什么权限?

权限级别是执行命令。安装流程涉及命令执行,可能通过 npx skills add https://github.com/masanao-ohba/claude-manifests --skill workflow-patterns 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

workflow-patterns 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

workflow-patterns 用的人多吗?

公开数据是累计安装 233 次,近一周安装 10 次,下载 82 次,GitHub 2 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

workflow-patterns 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 需要执行命令,装到常用环境前要先评估影响面。
  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理开发类任务、又不想每次都手写提示词的人。

不太适合

  • 不能接受 Skill 在你机器上执行命令的人。
  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

开发类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
skill-vetter

技能审查员 · OpenClaw

8,779,018986操作浏览器MIT-0通过
ontology

本体论 · OpenClaw

6,636,960576只读MIT-0通过
entra-app-registration

内部应用程序注册 · Codex、Claude

6,631,872764接触敏感数据未明确通过
entra-app-registration

内部应用程序注册 · Codex、Claude

3,757,680201接触敏感数据未明确通过
nano-banana-pro

纳米香蕉专业版 · OpenClaw

3,526,898364接触敏感数据MIT-0有提醒
workflow-patterns

工作流程模式 · 本页

2332执行命令未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Workflow Patterns

A technology-agnostic skill for managing multi-agent workflows via team orchestration.

Core Patterns

Team Orchestration Pattern

pattern: team_orchestration
description: "Create team, define tasks with dependencies, spawn workers"
use_when:
  - "Non-trivial tasks (small/medium/large)"
  - "Multiple agents needed"
  - "Quality gates required"

implementation:
  1_create_team: "TeamCreate with descriptive name"
  2_create_tasks: "TaskCreate with blockedBy dependencies"
  3_spawn_workers: "Task tool with team_name to spawn teammates"
  4_monitor: "Watch TaskList and teammate messages"
  5_evaluate: "Task(deliverable-evaluator) when chain complete"
  6_cleanup: "SendMessage(shutdown_request) + TeamDelete"

Task Dependency Pattern

pattern: task_dependencies
description: "Declare execution order via blockedBy relationships"
use_when:
  - "Steps must execute in order"
  - "Some steps can run in parallel"

example:
  tasks:
    - id: 1
      subject: "Implement feature"
      blockedBy: []
    - id: 2
      subject: "Run tests"
      blockedBy: [1]
    - id: 3
      subject: "Review code"
      blockedBy: [1]
    - id: 4
      subject: "Evaluate deliverables"
      blockedBy: [2, 3]

  # Tasks 2 and 3 run in parallel after task 1
  # Task 4 waits for both 2 and 3

Iteration Pattern

pattern: iteration
description: "Retry via new tasks until success or max iterations"
use_when:
  - "Quality gates must pass"
  - "Debugging cycles needed"
  - "Refinement required"

implementation:
  on_test_failure:
    1: "Create debug task (test-failure-debugger)"
    2: "Create fix task (code-developer, blockedBy: debug)"
    3: "Create retest task (test-executor, blockedBy: fix)"

  on_evaluation_fail:
    1: "Create fix tasks from DE recommendations"
    2: "Create retest + re-review tasks"
    3: "Create re-evaluate task (blockedBy: retest, re-review)"

  exit_conditions:
    - "DE returns PASS"
    - "Max iterations (3) reached"
    - "User intervention requested"

Direct Delegation Pattern

pattern: direct_delegation
description: "Simple Task call without team overhead"
use_when:
  - "Trivial tasks"
  - "Single agent needed"
  - "No coordination required"

implementation:
  single_task: "Task(code-developer) with full context"
  no_team: true
  no_task_list: true

Workflow Templates

Standard Implementation Workflow

template: standard_implementation
scale: small|medium

phases:
  analysis:
    - Task(goal-clarifier) → acceptance_criteria
    - Task(task-scale-evaluator) → scale, subtasks

  execution:
    - TeamCreate("dev-workflow")
    - TaskCreate: implement (no deps)
    - TaskCreate: test (blockedBy: implement)
    - TaskCreate: review (blockedBy: implement)
    - TaskCreate: evaluate (blockedBy: test, review)
    - Spawn: code-developer, test-executor, quality-reviewer

  evaluation:
    - Task(deliverable-evaluator) when tasks complete
    - On FAIL: create fix tasks, loop
    - On PASS: cleanup team

  cleanup:
    - SendMessage(shutdown_request) to all teammates
    - TeamDelete

Full Feature Workflow

template: full_feature
scale: large

phases:
  analysis:
    - Task(goal-clarifier) → acceptance_criteria
    - Task(task-scale-evaluator) → scale, subtasks, parallel_groups

  design:
    - Task(design-architect) → architecture spec

  execution:
    - TeamCreate("dev-workflow")
    - TaskCreate per subtask with dependency graph
    - Spawn: code-developer, test-developer, test-executor, quality-reviewer

  evaluation:
    - Task(deliverable-evaluator)
    - Retry loop (max 3)

  cleanup:
    - Shutdown teammates, TeamDelete

Debug Loop

template: debug_loop
trigger: "Test failures during execution"

steps:
  1_analyze:
    - TaskCreate: "Debug test failures"
    - Assign to test-failure-debugger (or use Task delegation)

  2_fix:
    - TaskCreate: "Apply fix" (blockedBy: debug task)
    - Assign to code-developer

  3_verify:
    - TaskCreate: "Re-run tests" (blockedBy: fix task)
    - Assign to test-executor

  4_check:
    - If pass: continue to review/evaluate
    - If fail: loop (max 3 iterations)

Coordination Strategies

Progress Tracking via TaskList

tracking:
  use: "TaskList tool to check status"
  states:
    pending: "Not yet started"
    in_progress: "Being worked on"
    completed: "Done"

  monitoring:
    - "Check TaskList after teammate messages"
    - "Identify blocked tasks"
    - "Assign unblocked tasks to idle teammates"

Inter-Agent Communication

communication:
  teammate_updates:
    via: "SendMessage(type: message)"
    when: "Task completion, blocking issues, questions"

  team_announcements:
    via: "SendMessage(type: broadcast)"
    when: "Critical issues only (expensive)"

  shutdown:
    via: "SendMessage(type: shutdown_request)"
    when: "All work complete, DE returns PASS"

Error Handling

error_handling:
  test_failure:
    action: "Create debug + fix + retest tasks"
    max_retries: 3

  agent_failure:
    action: "Re-spawn teammate, reassign task"

  blocked_task:
    detection: "Task stuck in pending with unresolved blockedBy"
    action: "Investigate blocking task, escalate if needed"

  evaluation_fail:
    action: "Create targeted fix tasks from DE feedback"
    max_iterations: 3

Anti-Patterns

Over-Orchestration

anti_pattern: over_orchestration
description: "Using team for trivial tasks"
symptom: "TeamCreate for single-line change"
solution: "Use direct Task delegation for trivial work"

Missing Exit Conditions

anti_pattern: infinite_loop
description: "Retry without max iteration check"
symptom: "Endless fix-test-fail cycle"
solution: "Always enforce max 3 iterations, escalate to user"

Premature Evaluation

anti_pattern: premature_evaluation
description: "Calling DE before implementation chain completes"
symptom: "DE fails because tests haven't run"
solution: "Use blockedBy to ensure evaluate waits for test + review"

Integration

Used By

primary_users:
  - "/dev-workflow command": "Orchestration patterns"

secondary_users:
  - task-scale-evaluator: "Workflow recommendations"

平台分布

Codex

37.54%
按下载量换算31

Claude

31.92%
按下载量换算26

Cursor

18.79%
按下载量换算15

Gemini CLI

9.41%
按下载量换算8

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

执行命令

安装流程涉及命令执行,可能通过 npx skills add https://github.com/masanao-ohba/claude-manifests --skill workflow-patterns 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills