Token导航 LogoToken导航

Agent Skill · 研究检索

geist-learning-lab(精神学习实验室)

geist-learning-lab 用于记录任务执行中的错误、用户纠正、经验和能力缺口,适合在 Codex、Claude、Cursor、Gemini CLI 中希望让 Agent 持续沉淀问题、修正和最佳实践时使用。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。

geist-learning-lab 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,按安装量在研究检索类的 73,675 个 Skill 里排第 42,402。

总安装 1,077周安装 44GitHub Stars 14下载 345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:geist-learning-lab(精神学习实验室)
来源仓库:https://github.com/vercel-labs/skill-geist-learning-labs
仓库路径:skills/geist-learning-lab
安装命令:
npx skills add https://github.com/vercel-labs/skill-geist-learning-labs --skill geist-learning-lab
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/vercel-labs/skill-geist-learning-labs --skill geist-learning-lab

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。

用于记录任务执行中的错误、纠正和经验缺口。geist-learning-lab 属于研究检索类 Skill,可作为该场景下的辅助能力补充。适合让 Agent 持续沉淀问题、修正和最佳实践。可结合来源仓库和原始 README 进一步核验具体用法。安装方式:通过 npx skills add 命令从指定仓库添加技能。

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操作。

来源仓库内路径:skills/geist-learning-lab

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/vercel-labs/skill-geist-learning-labs --skill…;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

geist-learning-lab 怎么安装?

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

geist-learning-lab 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。

geist-learning-lab 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/vercel-labs/skill-geist-learning-labs --skill geist-learning-lab 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。

geist-learning-lab 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。

geist-learning-lab 用的人多吗?

公开数据是累计安装 1,077 次,近一周安装 44 次,下载 345 次,GitHub 14 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。

geist-learning-lab 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。
  • 需要处理研究检索类任务、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。

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
geist-learning-lab

精神学习实验室 · 本页

1,07714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

Geist Learning Lab

Build interactive learning experiences that teach by making the learner *do something, get feedback, and iterate* — using Geist's dark-first, minimal, precise UI language.

Design Philosophy

  • Learning loops, not pages — Every screen completes: prompt → attempt → feedback → refinement → checkpoint
  • Cognitive load management — Progressive disclosure, chunking, one concept + one action at a time
  • Error-friendly design — Mistakes are expected; the UI normalizes them and turns them into learning moments
  • Interactive representations — Sliders, toggles, diagrams, live code — not just text
  • Geist precision — Dark backgrounds, tight tracking, semantic color, 4px grid, Swiss typography

Critical Rules

  1. Every lesson must include at least one attempt+feedback loop — No passive-only pages
  2. Every interactive control must have a visible effect — Within ~100ms or show loading state
  3. Progress/review must exist even in a prototype — Track steps completed, checkpoints passed
  4. Active attempt before explanation — Ask a question or show a task, then explain
  5. Immediate, specific feedback — "Correct" isn't enough; explain *why*
  6. Multiple representations — Always at least 2 of: text, code, visual, interactive, real data
  7. Color is learning signal — Green=correct, Red=incorrect, Amber=hint/warning, Blue=info/definition
  8. Progressive disclosure by default — Depth behind "Why?", "Edge Cases", "Formal Definition"
  9. URL state for exploration — Shareable slider/toggle configurations via search params
  10. Misconception-first design — Bake in common traps: "Most people get this wrong because..."

The Learning Loop (Central Doctrine)

Every lesson repeats this cycle:

StepWhat HappensComponent
OrientWhat you'll learn in 5-10 minHeading + learning objective
AttemptLearner predicts/answers/edits firstQuickCheck, CodePlayground, ParameterDock
FeedbackImmediate + specific responseFeedback surface (green/red panel)
ExplainShort explanation anchored to their attemptCallout, WorkedExample
ExtendVariation, edge case, or transfer taskBeforeAfterSplit, ConceptExplorer
CheckpointQuick recall questionQuickCheck with ConfidenceRating
Reflect"What changed in your understanding?"Optional text input or self-rating

"Aha Moment" Design Patterns

PatternHow It WorksWhen to Use
Prediction Gap"What do you think happens if...?" then revealBefore introducing a concept
Contrast PairTwo near-identical cases with different outcomesClarifying subtle distinctions
Slider to FailureLet learner push a parameter until it breaksUnderstanding boundaries/limits
Time TravelScrub algorithm step-by-step, watch invariantsAlgorithms, state machines
Misconception TrapTempting wrong option, then explain why wrongCommon errors in a domain

Quick Color Reference (Learning Semantics)

SignalColor TokenHexUsage
Correctgreen-700#46A758Success feedback, completed steps
Incorrectred-700#E5484DError feedback, failed checks
Hint/Cautionamber-700#FFB224Hints, warnings, "watch out"
Info/Definitionblue-700#0070F3Definitions, current step, links
Defaultgray-400#737373Everything else: borders, muted text
Progress donegreen-700 dot#46A758Completed step indicator
Progress currentblue-700 ring#0070F3Current step focus state
Progress pendinggray-400 dot#737373Upcoming steps

Component Library (23 Patterns)

Lesson Flow & Progress

<LessonShell /> <ProgressRail /> <StepFlow /> <CheckpointCard /> <ResumeBanner />

Explorable Explanations

<ParameterDock /> <LiveOutputPanel /> <ConceptExplorer /> <InteractiveDiagram /> <BeforeAfterSplit /> <TimelineExplorer /> <ConceptMap />

Code Learning

<CodePlayground /> <DiffEditor /> <TaskRunner /> <SnippetCopy />

Quizzes & Active Recall

<QuickCheck /> <ConfidenceRating /> <HintLadder /> <MistakeAnalyzer />

Reading Support

<GlossaryPopover /> <Callout /> <WorkedExample />

Typography in Learning Context

ElementWhat It MeansStyle
Big heading"What you're doing"`text-heading-{32\24}` semibold, negative tracking
Small label"Where you are"`text-label-{14\12}` gray-400
Mono"What the computer sees"text-copy-14-mono or text-label-14-mono
BodyReading contenttext-copy-14 or text-copy-16, max-w-2xl

Layout Patterns

ContextMax WidthStructure
Reading contentmax-w-2xl / max-w-4xlSingle column, comfortable line length
Interactive labmax-w-6xlTwo panels: explanation + playground
Full explorermax-w-7xlThree zones: nav + content + lab panel

App Router Structure

/                                    → Landing / course picker
/learn                               → Course catalog + resume
/learn/[courseSlug]                   → Course overview, modules, progress
/learn/[courseSlug]/[moduleSlug]      → Module overview
/learn/[courseSlug]/[moduleSlug]/[lessonSlug]  → Lesson shell

File Organization

content/courses/<courseSlug>/course.json       → Metadata, module order
content/courses/<courseSlug>/<module>/<lesson>.mdx  → Lesson content
components/learning/                           → Reusable learning components
lib/learning/progress.ts                       → Schema + storage adapters
lib/learning/grading.ts                        → Quiz validation, code tests
lib/learning/spaced.ts                         → Review scheduling

References

  • references/lesson-shell.md — LessonShell, ProgressRail, StepFlow, CheckpointCard, ResumeBanner TSX patterns
  • references/explorable-explanations.md — ParameterDock, LiveOutputPanel, ConceptExplorer, InteractiveDiagram, BeforeAfterSplit, TimelineExplorer, ConceptMap
  • references/code-learning.md — CodePlayground (read-only/guided/free), DiffEditor, TaskRunner, SnippetCopy
  • references/quiz-assessment.md — QuickCheck, ConfidenceRating, HintLadder, MistakeAnalyzer with feedback patterns
  • references/reading-support.md — GlossaryPopover, Callout variants, WorkedExample with stepwise reveal
  • references/pedagogical-framework.md — The Learning Loop doctrine, progressive disclosure rules, aha patterns, scaffolded complexity
  • references/architecture.md — Next.js App Router routing, MDX content system, state management, file organization
  • references/progress-state.md — Progress schema, localStorage adapter, spaced repetition, URL state for explorations
  • references/learning-animations.md — Step transitions, feedback reveals, diagram animations, reduced-motion compliance

平台分布

Codex

35.58%
按下载量换算123

Claude

31.35%
按下载量换算108

Cursor

16.05%
按下载量换算55

Gemini CLI

8.29%
按下载量换算29

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/vercel-labs/skill-geist-learning-labs --skill geist-learning-lab 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