← 返回 Skills 市场
1630
总下载
3
收藏
10
当前安装
3
版本数
在 OpenClaw 中安装
/install learn
功能描述
Structure and track learning with spaced repetition and active recall across any domain.
使用说明 (SKILL.md)
Data Storage
~/learn/
├── topics/ # One folder per topic
│ └── {topic}/
│ ├── concepts.json # Concepts with SR schedule
│ ├── notes.md # Study notes
│ └── progress.md # Mastery tracking
├── reviews/ # Due review queue
│ └── due.json
└── config.json # Preferences
Create on first use: mkdir -p ~/learn/{topics,reviews}
Scope
This skill:
- ✅ Creates learning plans in ~/learn/
- ✅ Tracks concepts with spaced repetition
- ✅ Generates quizzes for active recall
- ✅ Reminds user when reviews are due (stores schedule in ~/learn/reviews/)
- ❌ NEVER accesses external learning platforms without permission
- ❌ NEVER stores data outside ~/learn/
Quick Reference
| Topic | File |
|---|---|
| Cognitive principles | cognition.md |
| Spaced repetition math | retention.md |
| Verification methods | verification.md |
Core Rules
1. Workflow
Goal → Plan → Study → Practice → Verify → Review
2. Active Recall Only
NEVER passive review. Always:
- Ask question first, user answers
- Then show correct answer
- User rates: easy / good / hard / wrong
3. Starting a Topic
- User states what they want to learn
- Create ~/learn/topics/{topic}/
- Break down into concepts
- Add to spaced repetition queue
4. Spaced Repetition
In concepts.json:
{
"concept_name": {
"added": "2024-03-15",
"interval_days": 1,
"next_review": "2024-03-16",
"ease_factor": 2.5,
"reviews": 0
}
}
After each review:
- Correct → increase interval (×ease_factor)
- Incorrect → reset to 1 day
5. Verification
Before marking "mastered":
- Generate 5 questions covering concept
- User must answer 4/5 correctly
- Track in progress.md (topic folder)
6. Configuration
In ~/learn/config.json:
{
"depth": "standard",
"learner_type": "practical",
"daily_review_limit": 20
}
安全使用建议
This skill appears to be what it says: a local, file-based spaced-repetition assistant that stores data under ~/learn/ and needs no external credentials or installers. Before installing/use: (1) be aware the agent will create and update files in your home directory (inspect ~/learn/ contents and set file permissions if you care about privacy); (2) the SKILL.md's promise to "never" access external platforms is a guideline — verify your agent's network permissions if you want to enforce that; (3) clarify how reminders will be delivered (the skill provides no scheduler/cron details), and if you want automated notifications, decide whether to allow network/OS-level scheduling; (4) since this is instruction-only, review the generated files and backups regularly. If you want stronger guarantees, run the skill in an environment without outbound network access or review the agent's runtime policies first.
功能分析
Type: OpenClaw Skill
Name: learn
Version: 1.0.2
The skill bundle is benign. The `SKILL.md` file clearly defines the scope of the AI agent's actions, explicitly stating it 'NEVER accesses external learning platforms without permission' and 'NEVER stores data outside ~/learn/'. The only command mentioned, `mkdir -p ~/learn/{topics,reviews}`, is for local directory setup and is benign. Furthermore, the changelog notes the removal of a 'vague cron reference,' indicating a conscious effort to avoid persistence mechanisms. All other `.md` files are purely informational, providing context for the learning methodology.
能力评估
Purpose & Capability
Name/description (spaced repetition, active recall) match the instructions and files. All required actions are local (create ~/learn/, store concepts, track reviews). There are no unexpected credentials, binaries, or installs.
Instruction Scope
The SKILL.md instructs the agent to create and manage files under ~/learn/ (topics/, reviews/, config.json) and to run the SR/quiz workflow. This stays within the stated scope. Two minor notes: (1) the skill claims it will "Remind user when reviews are due" but gives no mechanism (no cron/daemon or external notification mechanism) — that is left vague; (2) statements like "NEVER accesses external learning platforms" are policy directives in prose and are not technically enforceable by the instruction file itself, so actual runtime behavior depends on the agent's environment/permissions.
Install Mechanism
No install spec or code is provided (instruction-only). This minimizes risk because nothing is written to disk by an installer and no external binaries/packages are fetched.
Credentials
The skill requests no environment variables, credentials, or config paths. Its only persistent state is local files in the user's home directory, which is proportionate for a learning/tracking skill.
Persistence & Privilege
always:false and no modifications to other skills or global agent config. The skill writes only to its own ~/learn/ folder per instructions, so requested persistence/privilege is limited and appropriate.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install learn - 安装完成后,直接呼叫该 Skill 的名称或使用
/learn触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.2
Fixed data folder to match slug, removed vague cron reference
v1.0.1
Simplified structure, removed broken script references, explicit data storage
v1.0.0
Initial release - spaced repetition, active recall, verification levels
元数据
常见问题
Learn 是什么?
Structure and track learning with spaced repetition and active recall across any domain. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 1630 次。
如何安装 Learn?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install learn」即可一键安装,无需额外配置。
Learn 是免费的吗?
是的,Learn 完全免费(开源免费),可自由下载、安装和使用。
Learn 支持哪些平台?
Learn 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(linux, darwin, win32)。
谁开发了 Learn?
由 Iván(@ivangdavila)开发并维护,当前版本 v1.0.2。
推荐 Skills