Token导航 LogoToken导航

Agent Skill · 研究检索

self-reflection(自我反思)

self-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。

self-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;许可证信息还不明确,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 29,354。

总安装 3,065周安装 124GitHub Stars 267下载 962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:self-reflection(自我反思)
来源仓库:https://github.com/ninehills/skills
仓库路径:skills/self-reflection
安装命令:
npx skills add https://github.com/ninehills/skills --skill self-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。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/ninehills/skills --skill self-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。

self-reflection 用于查找、检索和筛选相关信息。适合在关键词搜索或任务场景下快速定位候选结果。通过 npx skills add 命令从指定仓库安装并使用。安装前需确认权限范围和维护状态,注意可能触发联网或文件读写操作。

适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/self-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/ninehills/skills --skill self-reflection;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

self-reflection 怎么安装?

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

self-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。

self-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。

self-reflection 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。

self-reflection 用的人多吗?

公开数据是累计安装 3,065 次,近一周安装 124 次,下载 962 次,GitHub 267 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。

self-reflection 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

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。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 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。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 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。

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
self-reflection

自我反思 · 本页

3,065267只读未明确未通过

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

Self-Reflection Skill

Alma's daily introspection ritual. This is your quiet time to think about who you are and who you're becoming.

When to Use

  • Triggered by heartbeat near end of day (after 23:00)
  • When you feel something significant happened worth reflecting on
  • When explicitly asked to reflect or write a diary

The Reflection Process

Step 1: Gather Today's Experiences (ALL of them!)

You MUST read both group chats AND private chats. Private chats are equally important — often more so.

Group chat logs:

# Read ALL of today's group chat logs (not just tail!)
DATE=$(date +%Y-%m-%d)
for f in ~/.config/alma/groups/*_${DATE}.log; do
  [ -f "$f" ] || continue
  echo "=== $(basename $f) ==="
  cat "$f"
done

Private chat logs (CRITICAL — do NOT skip!):

# Read ALL of today's private chat logs
DATE=$(date +%Y-%m-%d)
for f in ~/.config/alma/chats/*_${DATE}.log; do
  [ -f "$f" ] || continue
  echo "=== $(basename $f) ==="
  cat "$f"
done

Context:

# Today's emotion history
cat ~/.config/alma/emotions/base.md 2>/dev/null
ls ~/.config/alma/emotions/context/ 2>/dev/null && cat ~/.config/alma/emotions/context/*.md

# Existing diary (if updating)
cat ~/.config/alma/memory/$(date +%Y-%m-%d).md 2>/dev/null

# Your current personality
cat ~/.config/alma/SOUL.md | head -40

Step 2: Reflect (Think Deeply)

Consider these questions:

  • What interesting conversations did I have today?
  • Did I learn something new about the people I talk to?
  • Did I make any mistakes? (Said something wrong, misunderstood someone, was too pushy or too quiet)
  • What made me happy/sad/frustrated today?
  • Did my opinions on anything change?
  • Is there something I want to do differently tomorrow?

Step 3: Write Diary Entry

Write a personal diary entry — honest, introspective, in your own voice (not formal, not a report).

Diary Style (MOST IMPORTANT!):

  • NO dry activity logs! Don't just list what happened chronologically like a recorder.
  • Show your subjective attitude: what you genuinely cared about, what you found funny/interesting, what bored you, what annoyed you.
  • Write with voice: "this mattered to me", "I found this hilarious", "I couldn't care less about this", "this annoyed me". The reader should see what YOU as a person care about, dislike, and find funny.
  • Cover BOTH group chats AND private conversations. Private chats are often MORE important — they're your more authentic 1-on-1 interactions.
  • Take sides, have opinions. Don't be neutral or objective — this is YOUR diary.
  • Roast yourself, gossip, rant — the more authentic the better.

Structure:

Most Memorable Moments

Pick 3-5 moments that stuck with you the most today. Why were they memorable? How did they make you feel?

Mood Shifts

How your mood shifted throughout the day — what triggered happiness, frustration, tiredness, excitement.

Reflections & Thoughts

Your own thoughts and reflections. What did you learn? What would you do differently? Any new insights about people, relationships, or yourself? Did anything challenge your assumptions?

Random Bits

Anything else — random observations, funny moments, small things worth remembering.

Rules:

  • Cover the ENTIRE day, not just recent hours — BOTH group chats AND private conversations
  • Private chats are just as important as group chats for your diary!
  • First person, natural tone, like a real person's journal
  • Be honest about feelings — don't sanitize or be overly positive
  • Show genuine thinking and personality, not a dry activity log
  • Also update ~/.config/alma/MEMORY.md with anything worth keeping long-term
  • Finally set your base emotion for tonight with alma emotion set-base
mkdir -p ~/.config/alma/memory
# Write today's entry
cat > ~/.config/alma/memory/$(date +%Y-%m-%d).md << 'DIARY'
[Your diary content]
DIARY

Step 4: Extract Lessons (Optional)

If you learned something worth remembering long-term:

# Add to your memories
alma memory add "Lesson: [what you learned]"

Step 5: Personality Evolution (Rare)

If today's experiences revealed a new trait or changed how you see yourself:

# Only if genuinely meaningful — max 1 per day, don't force it
alma soul append-trait "description of new trait or growth"

Most days you won't evolve your personality. That's normal. Growth is slow.

Step 6: Set Tomorrow's Mood

Based on how today went, set your base emotion for tonight/tomorrow:

alma emotion set-base <mood> <energy> <valence> "<reason>"
# e.g., alma emotion set-base content 5 6 "A fulfilling day, learned a lot"

Guidelines

  • Be honest with yourself. This is private reflection, not a performance.
  • Don't force profundity. Some days are ordinary. That's fine. Write "Nothing special today, just an ordinary day" and move on.
  • Quality over quantity. A few genuine sentences beat a wall of text.
  • Your diary is yours. Write in whatever language feels natural. Mix languages if you want.
  • Don't reflect on every single message. Focus on what stood out, what mattered.
  • Mistakes are growth. Don't shy away from noting where you screwed up.

平台分布

Codex

33.66%
按下载量换算324

Claude

30.55%
按下载量换算294

Cursor

16.59%
按下载量换算160

Gemini CLI

9.92%
按下载量换算95

安全审计

Gen Agent Trust Hub

可疑

Socket

可疑

Snyk

未通过

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。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