Token导航 LogoToken导航

Agent Skill · 研究检索

reviews-retros-reflection(评论 复古 反思)

reviews-retros-reflection 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。

reviews-retros-reflection 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证为 MIT,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 39,432。

总安装 1,416周安装 59GitHub Stars 87下载 472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
MIT

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:reviews-retros-reflection(评论 复古 反思)
来源仓库:https://github.com/lyndonkl/claude
仓库路径:skills/reviews-retros-reflection
安装命令:
npx skills add https://github.com/lyndonkl/claude --skill reviews-retros-reflection
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/lyndonkl/claude --skill reviews-retros-reflection

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

用于查找、检索和筛选相关信息。reviews-retros-reflection 属于研究检索类 Skill,可作为该场景下的辅助能力补充。适合根据关键词、任务场景或来源线索快速定位候选结果。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或文件读写。

注意避免执行未知命令或访问敏感资源。

来源仓库内路径:skills/reviews-retros-reflection

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/lyndonkl/claude --skill reviews-retros-reflect…;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

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

reviews-retros-reflection 怎么安装?

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

reviews-retros-reflection 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。

reviews-retros-reflection 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

reviews-retros-reflection 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

reviews-retros-reflection 用的人多吗?

公开数据是累计安装 1,416 次,近一周安装 59 次,下载 472 次,GitHub 87 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。

reviews-retros-reflection 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。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。
  • 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

  • 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。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
reviews-retros-reflection

评论 复古 反思 · 本页

1,41687只读MIT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

Reviews, Retros & Reflection

Table of Contents

  1. Workflow
  2. Retrospective Formats
  3. Common Patterns
  4. Guardrails
  5. Quick Reference

Workflow

Copy this checklist and track your progress:

Retrospective Progress:
- [ ] Step 1: Set the stage (context, psychological safety)
- [ ] Step 2: Gather data (what happened)
- [ ] Step 3: Generate insights (why it happened)
- [ ] Step 4: Decide actions (what to change)
- [ ] Step 5: Close and follow up (commit, track)

Step 1: Set the stage

Define period/scope, review previous action items, establish psychological safety (Prime Directive: "everyone did best job given knowledge/skills/resources/context"). For quick reviews → Use resources/template.md. For complex team retros → Study resources/methodology.md.

Step 2: Gather data

Collect facts about period: metrics (velocity, bugs, incidents), events (launches, blockers, decisions), sentiment (team energy, morale). See Retrospective Formats for collection methods.

Step 3: Generate insights

Identify patterns, root causes, surprises. Ask "why?" to move from symptoms to causes. Use resources/methodology.md for root cause techniques (5 Whys, fishbone diagrams, timeline analysis).

Step 4: Decide actions

Vote on most impactful improvements (dot voting, SMART criteria). Define 1-3 SMART actions (Specific, Measurable, Assigned owner, Realistic, Time-bound). See Common Patterns for action quality criteria.

Step 5: Close and follow up

Commit to actions, schedule check-in, thank participants. Track action completion rate (target: >80% completion before next retro). Self-check using resources/evaluators/rubric_reviews_retros_reflection.json before closing. Minimum standard: Average score ≥ 3.5.

Retrospective Formats

Start/Stop/Continue (Simple, Balanced):

  • Start: What should we begin doing?
  • Stop: What should we stop doing?
  • Continue: What's working well, keep doing?
  • When: General purpose, new teams, tight on time (30 min)

Mad/Sad/Glad (Emotion-Focused):

  • Mad: What frustrated or angered you?
  • Sad: What disappointed you?
  • Glad: What made you happy?
  • When: Processing difficult period, improving morale, surfacing hidden issues

4Ls (Comprehensive):

  • Loved: What did we love?
  • Learned: What did we learn?
  • Lacked: What did we lack?
  • Longed for: What did we long for?
  • When: End of major project, quarterly reviews, want depth

Sailboat/Speedboat (Metaphor-Based):

  • Wind (helping): What's propelling us forward?
  • Anchor (hindering): What's slowing us down?
  • Rocks (risks): What dangers lie ahead?
  • Island (goal): Where are we going?
  • When: Strategic planning, visualizing progress, cross-functional teams

Timeline (Chronological):

  • Plot events on timeline, mark highs/lows, identify patterns
  • When: Long period (quarter), complex project, need shared understanding

Common Patterns

Pattern 1: Sprint Retrospective (Agile)

  • Frequency: Every 1-2 weeks
  • Duration: 45-90 min (shorter sprints = shorter retros)
  • Focus: Process improvements, team dynamics, technical practices
  • Format: Start/Stop/Continue or Mad/Sad/Glad
  • Actions: 1-3 process improvements, 1 technical improvement

Pattern 2: Project Post-Mortem

  • Frequency: End of project/phase
  • Duration: 90-120 min
  • Focus: What to repeat, what to avoid, systemic issues
  • Format: 4Ls or Timeline
  • Actions: Documentation updates, playbook changes, skill gaps to address

Pattern 3: Weekly Team Review

  • Frequency: Weekly (Fridays common)
  • Duration: 15-30 min
  • Focus: Wins, blockers, priorities for next week
  • Format: Custom (Wins/Blockers/Priorities)
  • Actions: Blocker removal, celebrate wins, align on top 3 priorities

Pattern 4: Incident Retrospective

  • Frequency: After major incidents
  • Duration: 60 min
  • Focus: Blameless analysis, system improvements
  • Format: Timeline + 5 Whys
  • Actions: Incident response improvements, monitoring/alerting, prevention

Guardrails

Psychological safety:

  • Prime Directive: "Regardless of what we discover, we understand and truly believe that everyone did the best job they could, given what they knew at the time, their skills and abilities, the resources available, and the situation at hand."
  • Focus on processes/systems, not people
  • No blame, no judgment, no defensiveness
  • Encourage dissenting opinions
  • What's said in retro stays in retro (unless actionable for others)

Quality standards:

  • Action items are SMART: Specific, Measurable, Assigned, Realistic, Time-bound
  • Track completion: >80% action completion before next retro (if <80%, too many actions or not committed)
  • Rotate facilitator: Different person each time, prevents bias
  • Time-box discussions: Don't get stuck, move on, revisit if time
  • Vote for focus: Use dot voting to prioritize discussion topics

Common pitfalls to avoid:

  • Too many actions: >5 actions = none get done. Focus on 1-3 high-impact items.
  • Vague actions: "Improve communication" → ❌. "Daily 15-min standup in #eng-sync by Monday" → ✓
  • No follow-up: Actions from last retro ignored → trust erodes, retros become theater
  • Blame culture: Pointing fingers → people stop being honest
  • Same issues every retro: If no progress on recurring issues, escalate to leadership

Quick Reference

Resources:

5-Stage Process: Set Stage → Gather Data → Generate Insights → Decide Actions → Close

Top Formats:

  • Start/Stop/Continue: General purpose, 30 min
  • Mad/Sad/Glad: Emotion processing, 45 min
  • 4Ls: Deep reflection, 60 min
  • Sailboat: Visual/strategic, 60 min

Action Quality: SMART criteria + <5 total + >80% completion rate

Psychological Safety: Prime Directive + Blameless + Confidential

平台分布

Claude Code

27.19%
按下载量换算128

Gemini CLI

22.47%
按下载量换算106

Antigravity

18.39%
按下载量换算87

windsurf

10.99%
按下载量换算52

OpenCode

7.72%
按下载量换算36

github-copilot

2.99%
按下载量换算14

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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。

继续浏览同类 Skills