SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsli-index

li-index

General & other

View repository

Updated 10-15 23:32

About this skill

Material Library Index Maintenance

Reading Notes Library / Core Concepts Library / Viral Script Library / 04-Methodology Accumulation, as well as any libraries created in the future.

Two modes:

  • Incremental Maintenance Mode(default): the user reports "added a file", append an index entry for a single file
  • Health Check Mode: the user says "index health check", scan the entire library ↔ INDEX comparison, batch-fill missing entries, and clean up invalid entries

Mode Determination

用户触发词 → 判断模式：
- "加了 X" / "把这个放进索引" / "更新索引" → 增量维护模式
- "索引体检" / "检查索引" / "索引同步" / "补齐索引" → 体检模式

Incremental Maintenance Mode

General Workflow

Step 1: Confirm the target library and the new file path

Determine which library it is based on the user's description, and use Glob under 内容素材库/ to locate the file. If the path is complete, proceed directly to the next step.

Step 2: Read the full original text of the new file

Use the Read tool to read the file'sentire content, do not read only fragments.

This is the core step: the applicable scenarios and keywords of the index entry must come from the original text. Inferring from the file name/title is almost certainly inaccurate—when the AI later uses the index for retrieval, it will fail to match, or match content that does not correspond.

After reading, extract:

  • Core Topic: summarize in one sentence what this content is about
  • Applicable Scenarios: what types of topics/content will use it (specific to the scenario, such as "talking about procrastination" or "talking about product storytelling")
  • Keywords: 5-8, prioritizing counterintuitive concepts with specific operations and unique perspectives

Keyword Quality Assessment:

✅ Good Keywords ❌ Poor Keywords
Painkiller vs Vitamin (has comparative judgment) Entrepreneurship Methods (too general)
Dopamine is anticipation, not pleasure (overturns common sense) Psychological Knowledge (too broad)
If-Then Planning (has specific operations) Goal-Setting Techniques (no distinctive features)

Step 3: Read the target library's index file and understand the format

Read the corresponding library's index file and observe:

  • What is the table column structure?
  • Are the sections divided by topic, by workflow stage, or by content type?
  • How are file paths written? Are there special characters?
  • Is there a decision rules section?

New entries must strictly follow the existing format; do not change the column structure or path notation.

Step 4: Write to the index

Append the new entry at the end of the corresponding section, then determine whether the decision rules need to be updated:

  • The new content covers a topic type that the rulesdo not cover→ add a new rule
  • The new content is an effective supplement to a certain direction → append "or see 『file name』"
  • Highly overlapping with existing entries → skip

Special Notes for Each Library

Reading Notes Library

内容素材库/读书笔记库/INDEX.md

  • Section Logic: Partition by theme (personal growth/psychology/communication/entrepreneurship/philosophy/technology/creation)
  • Special requirements: Book-type files are usually very long, and the full text must be read; you cannot just read the first few paragraphs; the core viewpoints of a book are often in the middle and later sections
  • Path note: Book titles may contain multiple spaces (e.g. 《黑客与画家》 保罗·格雷厄姆.md has double spaces), and the path must match the actual file name

Core concept library

内容素材库/核心概念库/INDEX.md

  • Partition logic: Partition by workflow stage (topic selection stage/script writing stage/title and cover stage); each file represents a set of methodologies or templates
  • Keyword direction: Focus on method framework names (such as "viral formula" and "cognitive hijacking"), rather than specific viewpoints in the book
  • Decision rules: Rules must be linked to specific creation steps (which one to read when deepening the topic, which one to read when generating the script)

Viral script library

内容素材库/爆款脚本库/爆款索引.md

  • Different trigger timing: This index is only updated whenthe content's post-publication data reaches the viral standard, not when new files are added
    • Balanced viral hit: both likes and saves are 2000+
    • Save-type viral hit: save ratio > 1.5
    • Follower-growth viral hit: follower growth rate > 2%
  • Completely different format: Each entry is a structured item, containing data/type/file path/reusable structure/punchline; there is no table and no decision rules
  • Writing format(refer to existing entries): 
    #### [内容标题]
- **数据**：点赞X、收藏X（收藏比X）、涨粉X（涨粉率X%）
- **类型**：[内容类型，如方法论型/工具教程型/经历型]
- **路径**：`[文件路径]`
- **可复用结构**：
  - 开场：[开场方式]
  - 主体：[主体结构]
  - 结尾：[结尾方式]
- **金句**："[最具代表性的一句话]"

Build a new library's INDEX.md from scratch

If the target library does not yet have an index file, the process expands to:

  1. Use Glob to find all files under that library
  2. Read each file completely one by one (cannot skip; batch parallel reading is allowed)
  3. Refer to 内容素材库/核心概念库/INDEX.md 's format to create a new file, and add AI usage rules at the beginning
  4. After it is created, update li-topic/SKILL.md, li-writer/SKILL.md, and CLAUDE.md so that the AI creation process knows the new index exists

Health check mode

Trigger words: "index health check", "check index", "index sync", "fill in index", "index health check"

Libraries subject to health check (library registry)

Library root directory INDEX file Scan method
内容素材库/读书笔记库/ INDEX.md File existence: Glob under the library *.md(excluding INDEX.md) ↔ paths already recorded in INDEX
内容素材库/核心概念库/ INDEX.md File existence: Glob under the library *.md(excluding INDEX.md) ↔ paths already recorded in INDEX
04-方法论沉淀/ INDEX.md File existence: Recursively Glob all *.md(excluding INDEX.md) ↔ paths already recorded in INDEX
05-选题研究/ INDEX.md File existence: Validate the two sections "viral topic analysis report" and "benchmark bloggers" separately; only when benchmark bloggers total ≥ 5 entries is it considered necessary to list blogger rows separately
内容素材库/爆款脚本库/ 爆款索引.md Data threshold: Cross-check rows in 03-数据统计/数据统计表.md that reach the viral standard ↔ titles already recorded in the viral index. Missing entry = data meets the standard but the index has not included it; invalid = the file path in the index no longer exists

Health check process

Step 1: Scan the entire library

Execute in parallel for each library:

  • Use Glob to list the actual files under the library
  • Read the corresponding INDEX.md, and extract the recorded relative paths from the table rows/links
  • Compute two sets:
    • 漏录 = actually exists - recorded in INDEX
    • 失效 = recorded in INDEX - actually exists

Step 2: Output the health check report

Show the diff to the user first,do not act directly：

📋 索引体检报告

📚 读书笔记库
  漏录（X 个）：
    - [文件路径]
  失效（X 个）：
    - [INDEX 中存在但文件已不在的条目]

📚 核心概念库
  ...

📚 04-方法论沉淀
  ...

总计：漏录 X 条，失效 X 条

Step 3: Wait for user confirmation

"The above is the diff. Do you want to fill in all of them? Or pick a few? For invalid entries, should they be deleted or marked first?"

Only proceed to Step 4 after receiving instructions.Do not default to fully automatic。

Step 4: Fill in missing records

For each file the user confirms to add, process it according to the [Incremental Maintenance Mode] workflow: fully Read the original text → extract topic/scenario/keywords → append to the correct section of the corresponding library INDEX.

All original texts can be Read in parallel, but writing to INDEX must be serial (to avoid edit conflicts).

Special handling for the viral script library：

  • Missing entries usually come from 02-已发布内容/ rather than the material library itself
  • When writing to 爆款索引.md , use the existing structured format (data/type/path/reusable structure/punchlines), do not apply the generic INDEX table format
  • The original path should be the real path of the published content, not the path within the script library

Step 5: Handle invalid entries

  • User chooses "delete" → delete the corresponding row from the INDEX table/links
  • User chooses "mark" → append at the end of the entry ⚠️ 文件已不存在，待确认

Step 6: Output the health check completion report

✅ 体检完成
  补齐：X 条 → [列出新增条目]
  清理：X 条
  保留待人工处理：X 条

Health check mode notes

  1. Path notation must be consistent with INDEX: some INDEXes use 相对路径.md, some use [名](路径), identify the existing format
  2. Same name with different path does not count as missing: for example, INDEX says 《X》.md but the actual file is 《X》 作者.md(with double spaces), this counts as a path error rather than a missing entry, and the user should be Asked first
  3. Do not read the viral script library: explained above
  4. For large libraries, prioritize parallel Read: methodology accumulation may have a dozen or so files, read them all in parallel before starting to write