AgentSkillsCN

learnings

按主题分类展示标记为深度学习资源的内容。当用户想要系统性地研读已保存的资料时,这一功能大显身手。

SKILL.md
--- frontmatter
name: learnings
description: Show resources marked for deep learning, grouped by topic. Use when the user wants to study saved content.
argument-hint: "[--topic TOPIC] [--progress]"

learnings

Show resources marked for deep study, organized by topic.

Usage

code
/learnings
/learnings --topic ai-ml
/learnings --progress

Options

  • --topic TOPIC - Show only a specific topic's learning resources in detail
  • --progress - Show reading progress (read vs unread learning resources)

Instructions

When the user invokes /learnings, follow these steps:

Step 0: Load Configuration

Read config.json from the engram project root to get the vault path:

json
{
  "vault_path": "~/Documents/Obsidian/WebCapture"
}

Use the vault_path value as {VAULT_PATH} in all paths below. Expand ~ to the user's home directory.

Step 1: Read Index

Read {VAULT_PATH}/_system/index.json

Step 2: Filter Learning Resources

Filter notes where deep_learn == true.

Step 3: Group by Topic

Group filtered notes by topic field.

Step 4: Display Overview

code
## Learning Resources (X total)

### {Topic Name} ({count} resources)
- [{title}]({url}) - {summary} [{read ? "Read" : "Unread"}]
  - Connections: {list related note titles that are also learning resources}
- [{title2}]({url2}) - ...

### {Topic Name 2} ({count} resources)
...

### Suggested Reading Order
Based on capture dates and connections:
1. Start with: {foundational article}
2. Then read: {article building on #1}
3. Deep dive: {most complex article}

### Knowledge Gaps
{Identify areas where the user has few resources. E.g., "You have 3 articles on agent orchestration but none on agent evaluation/testing."}

Step 5: Topic Detail Mode

If --topic TOPIC is specified:

  • Show only that topic's learning resources
  • Include full summaries and key points for each
  • Show connections between them as a learning path
  • Suggest related resources in other topics

Step 6: Progress Mode

If --progress is specified:

  • Show read vs unread counts per topic
  • Highlight items that have been in the queue longest
  • Suggest next items to read

Step 7: Mark as Read (if user requests)

When the user says they've read/studied a learning resource:

bash
obsidian-cli frontmatter "{note-path}" --edit --key "read" --value "true"

Then update index.json to set read: true, decrement stats.unread, and regenerate views/unread.md.

Notes

  • Vault: {VAULT_PATH}/
  • Index: _system/index.json
  • Learning resources are notes with deep_learn: true in the index
  • obsidian-cli frontmatter --edit handles YAML updates without manual parsing