context-mate by jezweb/claude-skills
npx skills add https://github.com/jezweb/claude-skills --skill context-mate一个与 Claude Code 自然工作流程协同的工具包。用其有益之处,弃其无用之处。
当调用 context-mate 时,先分析项目,再推荐工具。
检查是否存在以下文件(使用 Glob 模式,先不读取内容):
| 文件/模式 | 指示 |
|---|---|
SESSION.md | 会话跟踪处于活动状态 |
IMPLEMENTATION_PHASES.md | 正在使用分阶段规划 |
PROJECT_BRIEF.md | 项目已探索/规划 |
CLAUDE.md 或 .claude/ |
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
| 存在 AI 上下文 |
.claude/rules/ | 存在修正规则 |
package.json 或 requirements.txt | 具有依赖项 |
tests/ 或 *.test.* | 具有测试基础设施 |
git status --short # 是否有未提交的更改?
git log --oneline -3 # 最近的提交信息?
项目阶段:
| 阶段 | 迹象 | 推荐 |
|---|---|---|
| 新项目 | 没有 CLAUDE.md,没有阶段文件 | /explore-idea 或 /plan-project |
| 活跃开发 | 存在 SESSION.md 或阶段文件 | /continue-session,开发者代理 |
| 维护模式 | 存在文档,没有 SESSION.md | 新工作用 /plan-feature,审计用 project-health |
| 会话中期 | 存在未提交的更改 + SESSION.md | 继续当前工作,完成后使用 /wrap-session |
告知用户:
/continue-session 以恢复")test-runner 可用")示例:
项目分析
✓
CLAUDE.md- AI 上下文已配置 ✓SESSION.md- 会话跟踪处于活动状态(阶段 2 进行中) ✓.claude/rules/- 3 条修正规则 ○ 未检测到测试文件推荐:
- 运行
/continue-session以恢复阶段 2 的工作- 准备提交时使用
commit-helper代理- 添加测试时考虑使用
test-runner代理
保持在 10 行以内。不要信息过载 - 只突出显示相关内容。
这个名字有双重含义:
这不是"做事的正确方法" - 这些工具存在是因为上下文窗口是真实的限制,而不是因为我们规定方法论。
| 命令 | 功能 |
|---|---|
/context-mate | 分析项目,推荐工具 |
/explore-idea | 从一个模糊的想法开始 |
/plan-project | 规划一个新项目 |
/plan-feature | 规划一个特定功能 |
/wrap-session | 结束工作会话 |
/continue-session | 从上次会话恢复 |
/docs-init | 创建项目文档 |
/docs-update | 更改后更新文档 |
/brief | 在清除上下文前保存信息 |
/reflect | 捕获经验教训 → 规则、技能、记忆 |
/release | 准备部署 |
| 代理 | 功能 |
|---|---|
commit-helper | 编写提交信息 |
code-reviewer | 审查代码质量 |
debugger | 调查错误 |
test-runner | 运行/编写测试 |
build-verifier | 检查 dist 是否与源代码匹配 |
documentation-expert | 创建/更新文档 |
orchestrator | 协调多步骤工作 |
| 技能 | 提供内容 |
|---|---|
project-planning | 基于阶段的规划模板 |
project-session-management | SESSION.md 模式 |
docs-workflow | 文档维护命令 |
deep-debug | 多代理调试 |
project-health | AI 可读性审计 |
developer-toolbox | 上述 7 个代理 |
┌─────────────────────────────────────────────────────────────┐
│ 项目生命周期 │
├─────────────────────────────────────────────────────────────┤
│ /explore-idea → /plan-project → [工作] → /wrap-session │
│ ↓ ↓ ↓ ↓ │
│ PROJECT_BRIEF PHASES.md SESSION.md git 检查点 │
│ ↓ │
│ /continue-session │
│ ↓ │
│ [恢复工作] │
│ ↓ │
│ /reflect → /release │
└─────────────────────────────────────────────────────────────┘
| 您想要... | 使用这个 |
|---|---|
| 探索一个模糊的想法 | /explore-idea |
| 规划一个新项目 | /plan-project |
| 规划一个特定功能 | /plan-feature |
| 结束一个工作会话 | /wrap-session |
| 休息后恢复 | /continue-session |
| 创建/更新文档 | /docs-init, /docs-update |
| 调试顽固问题 | deep-debug 技能 |
| 审查代码质量 | code-reviewer 代理 |
| 使用 TDD 运行测试 | test-runner 代理 |
| 准备 git 提交 | commit-helper 代理 |
| 验证构建输出 | build-verifier 代理 |
| 检查文档是否对 AI 可读 | context-auditor 代理 |
| 验证工作流程是否有效 | workflow-validator 代理 |
| 检查会话交接质量 | handoff-checker 代理 |
用于完整项目生命周期的九个集成命令:
| 命令 | 目的 |
|---|---|
/explore-idea | 头脑风暴并验证项目概念 |
/plan-project | 生成分阶段实施计划 |
/plan-feature | 规划特定的功能添加 |
/docs-init | 创建初始项目文档 |
/docs-update | 更改后更新文档 |
/wrap-session | 以 git 检查点结束会话 |
/continue-session | 从 SESSION.md 恢复 |
/reflect | 审查进度并规划后续步骤 |
/release | 准备部署/发布 |
调用 : Skill(skill: "project-workflow")
使用带有 git 检查点的 SESSION.md 跨上下文窗口跟踪进度。
调用 : Skill(skill: "project-session-management")
七个用于常见开发任务的专用代理:
| 代理 | 用于 |
|---|---|
commit-helper | 生成有意义的提交信息 |
code-reviewer | 安全、质量、架构审查 |
debugger | 系统性错误调查 |
test-runner | TDD 工作流,测试创建 |
build-verifier | 验证 dist/ 是否与源代码匹配 |
documentation-expert | 创建/更新项目文档 |
orchestrator | 协调多步骤项目 |
调用 : Skill(skill: "developer-toolbox")
针对抵抗常规调试的顽固错误进行多代理调查。
调用 : Skill(skill: "deep-debug")
三个用于 AI 可读性和工作流质量的代理:
| 代理 | 目的 |
|---|---|
context-auditor | 检查文档是否对 AI 可读(评分 0-100) |
workflow-validator | 验证记录的流程是否有效(评分 0-100) |
handoff-checker | 验证会话连续性质量(评分 0-100) |
调用 : Skill(skill: "project-health")
四个用于文档管理的命令:
| 命令 | 目的 |
|---|---|
/docs | 快速文档查找 |
/docs-init | 创建初始文档 |
/docs-update | 更改后更新 |
/docs-claude | 生成 AI 优化的 CLAUDE.md |
调用 : Skill(skill: "docs-workflow")
会话 是上下文窗口(工作 2-4 小时后上下文会填满)。
阶段 是工作单元(如"阶段 1:数据库设置"这样的逻辑分组)。
一个阶段可能跨越多个会话。一个会话可能涉及多个阶段。它们是独立的概念。
Git 提交用作语义检查点,而不仅仅是版本控制:
# 不好:提交作为保存点
git commit -m "WIP"
git commit -m "更多更改"
# 好:提交作为进度标记
git commit -m "完成阶段 1:数据库模式和迁移"
git commit -m "阶段 2 部分:认证中间件工作正常,UI 待完成"
当通过 /continue-session 恢复时,这些提交会告诉您当前所处的位置。
技能会增量加载以保留上下文:
这意味着一个 50 项技能的工具包在您实际使用某项之前,只消耗约 2,500 个 token。
两个互补的知识系统:
| 技能 | 规则
---|---|---
位置 | ~/.claude/skills/ | .claude/rules/(项目内)
内容 | 丰富的捆绑包 | 单个 Markdown 文件
目的 | 教授如何使用 X | 修正过时的模式
示例 | 如何设置 Tailwind v4 | 修正 Claude 可能建议的 v3 语法
规则是项目可移植的 - 它们随仓库一起移动,因此任何 Claude 实例都能获得修正。
繁重的任务(代码审查、调试、测试)在子代理中运行,以便:
/explore-idea # 可选:澄清您要构建什么
/plan-project # 生成分阶段计划
# 处理阶段 1...
/wrap-session # 以检查点结束
/continue-session # 读取 SESSION.md,建议后续步骤
# 继续工作...
/wrap-session # 再次创建检查点
/plan-feature # 规划特定功能
# 实现...
/wrap-session # 创建检查点
# 如果常规调试无效:
Skill(skill: "deep-debug")
# 生成调查代理
上下文窗口是真实的。 它们会填满。工作会丢失。会话会结束。
这些工具不与之对抗 - 而是与之协同工作:
用其有益之处。弃其无用之处。
这是刀叉勺式项目管理学派:
| 传统项目管理 | Context Mate |
|---|---|
| "遵循方法论" | "一切都会好的" |
| "更新甘特图" | /wrap-session |
| "咨询 RACI 矩阵" | "喂 Claude,接下来做什么?" |
没有仪式。不与您的 AI 开站会。没有燃尽图。
如果荷马·辛普森在 30 秒内搞不明白,那就太复杂了。
这一切都与上下文有关,伙计。 🥄
每周安装数
142
仓库
GitHub 星标数
643
首次出现
2026年2月2日
安全审计
安装于
claude-code117
replit97
opencode95
gemini-cli90
codex83
cursor81
A toolkit that works with Claude Code's natural flow. Use what helps, ignore what doesn't.
When context-mate is invoked, analyze the project first before recommending tools.
Check for these files (use Glob, don't read contents yet):
| File/Pattern | Indicates |
|---|---|
SESSION.md | Session tracking active |
IMPLEMENTATION_PHASES.md | Phased planning in use |
PROJECT_BRIEF.md | Project explored/planned |
CLAUDE.md or .claude/ | AI context exists |
.claude/rules/ | Correction rules present |
package.json or requirements.txt | Has dependencies |
tests/ or *.test.* | Has test infrastructure |
git status --short # Uncommitted changes?
git log --oneline -3 # Recent commit messages?
Project Stages:
| Stage | Signs | Recommend |
|---|---|---|
| New Project | No CLAUDE.md, no phases | /explore-idea or /plan-project |
| Active Development | SESSION.md or phases exist | /continue-session, developer agents |
| Maintenance Mode | Docs exist, no SESSION.md | /plan-feature for new work, project-health for audits |
| Mid-Session | Uncommitted changes + SESSION.md |
Tell the user:
/continue-session to resume")test-runner available")Example:
Project Analysis
✓
CLAUDE.md- AI context configured ✓SESSION.md- Session tracking active (Phase 2 in progress) ✓.claude/rules/- 3 correction rules ○ No test files detectedRecommendations:
- Run
/continue-sessionto resume Phase 2 work- Use
commit-helperagent when ready to commit- Consider
test-runneragent when adding tests
Keep it under 10 lines. Don't overwhelm - just highlight what's relevant.
The name has a double meaning:
This isn't "The Correct Way To Do Things" - these tools exist because context windows are real constraints, not because we're dictating methodology.
| Command | What it does |
|---|---|
/context-mate | Analyze project, recommend tools |
/explore-idea | Start with a vague idea |
/plan-project | Plan a new project |
/plan-feature | Plan a specific feature |
/wrap-session | End work session |
/continue-session |
| Agent | What it does |
|---|---|
commit-helper | Writes commit messages |
code-reviewer | Reviews code quality |
debugger | Investigates bugs |
test-runner | Runs/writes tests |
build-verifier | Checks dist matches source |
documentation-expert | Creates/updates docs |
| Skill | What it provides |
|---|---|
project-planning | Phase-based planning templates |
project-session-management | SESSION.md patterns |
docs-workflow | Doc maintenance commands |
deep-debug | Multi-agent debugging |
project-health | AI-readability audits |
developer-toolbox | The 7 agents above |
┌─────────────────────────────────────────────────────────────┐
│ PROJECT LIFECYCLE │
├─────────────────────────────────────────────────────────────┤
│ /explore-idea → /plan-project → [work] → /wrap-session │
│ ↓ ↓ ↓ ↓ │
│ PROJECT_BRIEF PHASES.md SESSION.md git checkpoint │
│ ↓ │
│ /continue-session │
│ ↓ │
│ [resume work] │
│ ↓ │
│ /reflect → /release │
└─────────────────────────────────────────────────────────────┘
| You want to... | Use this |
|---|---|
| Explore a vague idea | /explore-idea |
| Plan a new project | /plan-project |
| Plan a specific feature | /plan-feature |
| End a work session | /wrap-session |
| Resume after a break | /continue-session |
| Create/update docs | /docs-init, |
Nine integrated commands for the complete project lifecycle:
| Command | Purpose |
|---|---|
/explore-idea | Brainstorm and validate project concepts |
/plan-project | Generate phased implementation plan |
/plan-feature | Plan a specific feature addition |
/docs-init | Create initial project documentation |
/docs-update | Update docs after changes |
/wrap-session | End session with git checkpoint |
Invoke : Skill(skill: "project-workflow")
Track progress across context windows using SESSION.md with git checkpoints.
Invoke : Skill(skill: "project-session-management")
Seven specialized agents for common development tasks:
| Agent | Use For |
|---|---|
commit-helper | Generate meaningful commit messages |
code-reviewer | Security, quality, architecture review |
debugger | Systematic bug investigation |
test-runner | TDD workflows, test creation |
build-verifier | Verify dist/ matches source |
documentation-expert | Create/update project docs |
Invoke : Skill(skill: "developer-toolbox")
Multi-agent investigation for stubborn bugs that resist normal debugging.
Invoke : Skill(skill: "deep-debug")
Three agents for AI-readability and workflow quality:
| Agent | Purpose |
|---|---|
context-auditor | Check if docs are AI-readable (score 0-100) |
workflow-validator | Verify documented processes work (score 0-100) |
handoff-checker | Validate session continuity quality (score 0-100) |
Invoke : Skill(skill: "project-health")
Four commands for documentation management:
| Command | Purpose |
|---|---|
/docs | Quick doc lookup |
/docs-init | Create initial docs |
/docs-update | Update after changes |
/docs-claude | Generate AI-optimized CLAUDE.md |
Invoke : Skill(skill: "docs-workflow")
Sessions are context windows (2-4 hours of work before context fills up).
Phases are work units (logical groupings like "Phase 1: Database Setup").
A phase might span multiple sessions. A session might touch multiple phases. They're independent concepts.
Git commits serve as semantic checkpoints , not just version control:
# Bad: commits as save points
git commit -m "WIP"
git commit -m "more changes"
# Good: commits as progress markers
git commit -m "Complete Phase 1: Database schema and migrations"
git commit -m "Phase 2 partial: Auth middleware working, UI pending"
When resuming via /continue-session, these commits tell the story of where you are.
Skills load incrementally to preserve context:
This means a 50-skill toolkit only costs ~2,500 tokens until you actually use something.
Two complementary knowledge systems:
| Skills | Rules
---|---|---
Location | ~/.claude/skills/ | .claude/rules/ (project)
Content | Rich bundles | Single markdown files
Purpose | Teach how to use X | Correct outdated patterns
Example | How to set up Tailwind v4 | Fix v3 syntax Claude might suggest
Rules are project-portable - they travel with the repo so any Claude instance gets the corrections.
Heavy tasks (code review, debugging, testing) run in sub-agents to:
/explore-idea # Optional: clarify what you're building
/plan-project # Generate phased plan
# Work on Phase 1...
/wrap-session # End with checkpoint
/continue-session # Reads SESSION.md, suggests next steps
# Continue working...
/wrap-session # Checkpoint again
/plan-feature # Plan the specific feature
# Implement...
/wrap-session # Checkpoint
# If normal debugging isn't working:
Skill(skill: "deep-debug")
# Spawns investigation agents
Context windows are real. They fill up. Work gets lost. Sessions end.
These tools don't fight that - they work with it:
Use what helps. Ignore what doesn't.
This is the knifey-spooney school of project management :
| Traditional PM | Context Mate |
|---|---|
| "Follow the methodology" | "She'll be right" |
| "Update the Gantt chart" | /wrap-session |
| "Consult the RACI matrix" | "Oi Claude, what next?" |
No ceremonies. No standups with your AI. No burndown charts.
If Homer Simpson can't figure it out in 30 seconds, it's too complicated.
It's all about the context, maaate. 🥄
Weekly Installs
142
Repository
GitHub Stars
643
First Seen
Feb 2, 2026
Security Audits
Gen Agent Trust HubPassSocketPassSnykPass
Installed on
claude-code117
replit97
opencode95
gemini-cli90
codex83
cursor81
AI Elements:基于shadcn/ui的AI原生应用组件库,快速构建对话界面
66,200 周安装
shadcn/ui 框架:基于 Tailwind CSS 的 React UI 组件库与设计系统
696 周安装
解决 Docker 沙盒 npm 安装崩溃:sandbox-npm-install 技能详解与使用指南
701 周安装
Himalaya 命令行电子邮件客户端 | 终端邮件管理工具 | IMAP/SMTP 配置指南
706 周安装
Sonos CLI 命令行工具:本地网络控制 Sonos 音箱,支持分组、收藏、Spotify 搜索
708 周安装
SwiftData 教程:iOS 26+ 数据持久化、查询与管理完整指南
709 周安装
Lightpanda 无头浏览器 - Zig 语言构建,AI 自动化与网页抓取利器,速度提升11倍
693 周安装
Continue current work, /wrap-session when done |
| Resume from last session |
/docs-init | Create project docs |
/docs-update | Update docs after changes |
/brief | Preserve context before clearing |
/reflect | Capture learnings → rules, skills, memory |
/release | Prepare for deployment |
orchestrator | Coordinates multi-step work |
/docs-update| Debug something stubborn | deep-debug skill |
| Review code quality | code-reviewer agent |
| Run tests with TDD | test-runner agent |
| Prepare a git commit | commit-helper agent |
| Verify build output | build-verifier agent |
| Check docs are AI-readable | context-auditor agent |
| Validate workflows work | workflow-validator agent |
| Check session handoff quality | handoff-checker agent |
/continue-session | Resume from SESSION.md |
/reflect | Review progress and plan next steps |
/release | Prepare for deployment/release |
orchestrator | Coordinate multi-step projects |