Token导航 LogoToken导航TokenDH.com
研究检索需要联网github未标认证来源可访问clear审计提醒

project-workflow项目工作流程

Agent Skill

project-workflow 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。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。

总安装

8,590

周安装

369

GitHub Stars

750

下载量

3,011
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

3

许可证

MIT

最后核验

2026-05-01

来源状态

来源可访问

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:project-workflow(项目工作流程)
来源仓库:https://github.com/jezweb/claude-skills
仓库路径:skills/project-workflow
安装命令:
npx skills add https://github.com/jezweb/claude-skills --skill project-workflow
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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/jezweb/claude-skills --skill project-workflow

简介

九个集成的斜线命令自动化项目生命周期从想法验证到发布。

  • 命令跨越探索、规划、执行、会话管理、上下文保存和发布阶段,每个项目总共节省 35-55 分钟的时间
  • /总结会话
  • 和/继续会话
  • 通过 SESSION.md 检查点和 git 提交管理多个会话的上下文
  • /释放
  • 在发布前执行八阶段安全检查,包括秘密扫描、文档验证和构建测试
  • /简介
  • 和/反映
  • 使用 /reflect 捕获未来会议的决策和操作知识
  • 建议定制代理以实现可重复的流程

SKILL.md

Project Workflow Skill

9 integrated slash commands for complete project lifecycle automation: idea validation → planning → execution → session management → context preservation → release.

Time savings: 35-55 minutes per project lifecycle

Installation

Marketplace: /plugin install project-workflow@jezweb-skills

Manual: Copy commands/*.md to ~/.claude/commands/

The 9 Commands

1. /explore-idea - Pre-Planning Exploration

Use when: Rough idea that needs tech stack validation, scope management, or research before planning.

Creates: PROJECT_BRIEF.md with validated decisions → hands off to /plan-project

Time savings: 10-15 min


2. /plan-project - Generate Project Planning Docs

Use when: Starting new project with clear requirements, or after /explore-idea.

Creates: IMPLEMENTATION_PHASES.md, SESSION.md, DATABASE_SCHEMA.md (if needed), API_ENDPOINTS.md (if needed), ARCHITECTURE.md

Invokes: project-planning skill

Time savings: 5-7 min


3. /plan-feature - Add Features to Existing Projects

Use when: Adding feature to existing project with SESSION.md + IMPLEMENTATION_PHASES.md.

Does: Generates new phases via project-planning skill, integrates into IMPLEMENTATION_PHASES.md with renumbering, updates SESSION.md.

Time savings: 7-10 min


4. /wrap-session - End-of-Session Checkpoint

Use when: Context full (>150k tokens), end of work session, or before task switch.

Does: Task agent analyzes session → updates SESSION.md (progress, Next Action, blockers) → git checkpoint commit → formatted handoff summary.

Time savings: 2-3 min


5. /continue-session - Start-of-Session Context Loading

Use when: Starting new session or after /wrap-session checkpoint.

Does: Explore agent loads SESSION.md + planning docs → shows git history + session summary (phase, progress, Next Action, blockers) → optionally opens file → asks permission to continue.

Time savings: 1-2 min


6. /workflow - Interactive Workflow Guide

Use when: First time user, unsure which command to use, or need quick reference.

Does: Shows all 9 commands → context-aware guidance with decision trees → offers to execute appropriate command.


7. /release - Pre-Release Safety Checks

Use when: Ready to push to public GitHub or create release.

8 Phases:

  1. Critical Safety (BLOCKERS): Secrets scan (gitleaks), personal artifacts check, git remote verification
  2. Documentation (REQUIRED): LICENSE, README (>100 words), CONTRIBUTING.md (>500 LOC), CODE_OF_CONDUCT (>1000 LOC)
  3. Configuration:.gitignore, package.json, git branch warning
  4. Quality (NON-BLOCKING): Build test, npm audit, large files (>1MB)
  5. Report: Blockers/warnings/recommendations + safe to release verdict 6-8. Auto-Fix & Publish: Fix issues, release prep commit, optional git tag + GitHub release

Time savings: 10-15 min


8. /brief - Context Preservation Document

Use when: Before clearing context, to preserve key decisions and requirements for future sessions.

Creates: docs/brief-[slug].md with extracted information from current conversation.

Does: Analyzes conversation → extracts what's being built, decisions made, technical details → creates organized markdown file → optionally creates GitHub issue for tracking.

Time savings: 3-5 min


9. /reflect - Capture Operational Knowledge

Use when: Before context compaction, after completing significant work, or when valuable learnings accumulated.

Does: Reviews conversation → identifies workflows discovered, patterns learned, tool sequences, corrections made → routes each learning to appropriate destination (rules, CLAUDE.md, skills, docs) → optionally suggests automation (scripts, commands, custom agents).

Key Feature: Considers whether processes should become custom agents when they require reasoning/decisions, benefit from parallel execution, or involve research/exploration.

Time savings: 5-10 min


Workflow Examples

Full: /explore-idea → /plan-project → work → /wrap-session → /continue-session → /plan-feature (if needed) → repeat → /reflect → /release

Quick (clear requirements): /plan-project → work → /wrap-session → /continue-session → /release

Context Preservation: /brief (before clearing context), /reflect (capture learnings)

Helpers: /workflow (guidance), /plan-feature (add feature), /release (publish)


Integration

project-planning: Invoked by /plan-project and /plan-feature (generates IMPLEMENTATION_PHASES.md, DATABASE_SCHEMA.md, API_ENDPOINTS.md)

project-session-management: SESSION.md protocol for /wrap-session and /continue-session

Claude Code agents: /wrap-session (Task agent), /continue-session + /explore-idea (Explore agent)


Command Relationships

EXPLORATION PHASE
/explore-idea (optional)
    ↓
    Creates PROJECT_BRIEF.md
    ↓
PLANNING PHASE
/plan-project (reads PROJECT_BRIEF.md if exists)
    ↓
    Creates IMPLEMENTATION_PHASES.md + SESSION.md
    ↓
EXECUTION PHASE
Work on phases
    ↓
/wrap-session (when context full)
    ↓
    Updates SESSION.md, git checkpoint
    ↓
/continue-session (new session)
    ↓
    Loads SESSION.md, continues work
    ↓
/plan-feature (when need new features)
    ↓
    Adds phases to IMPLEMENTATION_PHASES.md
    ↓
Continue wrap → resume cycle
    ↓
CONTEXT PRESERVATION
/brief (before clearing context)
    ↓
    Creates docs/brief-[slug].md
    ↓
/reflect (capture learnings)
    ↓
    Routes knowledge to rules, docs, CLAUDE.md
    ↓
RELEASE PHASE
/release (when ready to publish)
    ↓
    Safety checks → GitHub release

HELPER
/workflow (anytime)
    ↓
    Interactive guidance

Time Savings Breakdown

CommandTime SavedTasks Automated
/explore-idea10-15 minResearch, validation, scope management, tech stack evaluation
/plan-project5-7 minPlanning doc generation, git setup, phase structuring
/plan-feature7-10 minFeature planning, phase integration, doc updates
/wrap-session2-3 minSESSION.md updates, git checkpoint, handoff summary
/continue-session1-2 minContext loading, git history review, next action display
/workflowInstantNavigation, decision trees, command selection
/release10-15 minSecret scanning, doc validation, build testing, release creation
/brief3-5 minContext extraction, decisions capture, markdown generation
/reflect5-10 minLearning extraction, knowledge routing, automation suggestions

Total per project lifecycle: 45-70 minutes


Prerequisites

All: Claude Code CLI, git repo (recommended)

/plan-feature: Existing SESSION.md + IMPLEMENTATION_PHASES.md

/wrap-session, /continue-session: SESSION.md (created by /plan-project)

/release: Git repo with commits, package.json (Node.js), remote URL (for publishing)


Troubleshooting

/plan-project "No project description": Use /explore-idea first or discuss project with Claude

/plan-feature "Prerequisites not met": Run /plan-project first (creates SESSION.md + IMPLEMENTATION_PHASES.md)

/wrap-session "No git repository": Run git init

/continue-session "SESSION.md not found": Run /plan-project

/release "Secrets detected": Add to.gitignore, remove from git history


Version History

1.1.0 (2026-01-11)

  • Added /brief command for context preservation
  • Added /reflect command for capturing operational knowledge
  • /reflect now suggests custom agents for processes requiring reasoning
  • Total commands: 9

1.0.0 (2025-11-12)

  • Initial release
  • 7 integrated slash commands
  • Plugin marketplace distribution
  • Command bundling via plugin.json

Issues: https://github.com/jezweb/claude-skills/issues | Author: Jeremy Dawes (jeremy@jezweb.net) | License: MIT

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

04

需要参考平台分布和安装热度时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Claude Code

24.59%
按下载量换算740

Gemini CLI

23.96%
按下载量换算721

Antigravity

18.98%
按下载量换算571

Cursor

11.3%
按下载量换算340

OpenCode

8.1%
按下载量换算244

Codex

3.34%
按下载量换算101

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

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。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。

来源信息

继续浏览同类 Skills