Compile index

Skill chappyasel/meta-kb/.claude/skills/compile-index

A self-improving knowledge base about LLM agent infrastructure

Install

npx -y skills add chappyasel/meta-kb --skill compile-index

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
23 stars23 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Generates the agent-optimized ROOT.md topic index, project/topic indexes, missing coverage list, landscape comparison table, and wiki README. Use when generating indexes, the ROOT.md entry point, or the comparison table.

SKILL.md

3.2 KB, as published. Nobody here has run it

Compile Indexes

Read config/domain.ts for the domain name, topic, and taxonomy bucket definitions.

Read all compiled articles from wiki/ to understand what exists, then generate the 6 files below.

Prerequisites

All synthesis articles, reference cards, and field-map.md must exist in wiki/ before running this skill.

Output Files

1. wiki/ROOT.md

Agent-optimized topic index. MUST be under 2K tokens. Format:

Frontmatter: type: "root", version: 1, compiled_at, token_estimate, entities_total, sources_total
Topics section: one line per bucket with source count + brief description + path
Top Projects section: 10-12 highest-relevance projects with [bucket, stars, refs]: description -> path Star count accuracy: Only read star counts from raw/repos/ source files (the project's own repo metadata). Do NOT use star counts from papers, tweets, or articles that mention the project — those often contain another project's star count and cause data corruption.
Key Concepts section: 8-10 highest-relevance concepts with [bucket]: description -> path
Meta footer: compilation date, source/entity/edge counts, links to field-map and graph
Each entry is ONE LINE using compressed notation, not prose

See references/root-template.md for the exact format example.

2. wiki/indexes/projects.md

Markdown table with columns: Project | Bucket | Sources. Link each project name to its reference card (projects/{slug}.md).

3. wiki/indexes/topics.md

Bulleted list: concept name linked to its card (concepts/{slug}.md), followed by a one-line description.

4. wiki/indexes/missing.md

Entities mentioned in sources but lacking full articles. Columns: Name | Type (project/concept) | Bucket | Source Count. Sorted by source count descending.

5. wiki/comparisons/landscape.md

All-projects comparison table. Columns: Project | Bucket | Description (80 char max). Sorted by bucket, then alphabetically.

6. wiki/README.md

Human entry point for the wiki. Include:

What this knowledge base is (1-2 sentences, using name and topic from config/domain.ts)
Link to field-map.md
Table of synthesis articles (one per bucket, with name, title, word count)
Links to indexes (projects, topics, missing) and graph
Stats: source count, entity count, total word count

Frontmatter

Each file gets appropriate frontmatter:

# ROOT.md
---
type: root
version: 1
compiled_at: "YYYY-MM-DDTHH:MM:SS.000Z"
token_estimate: <actual count>
entities_total: <count>
sources_total: <count>
---

# Other index files
---
type: index
last_compiled: "YYYY-MM-DD"
---

Rules

ROOT.md is for agents. README.md is for humans. Keep them separate.
ROOT.md must stay under 2K tokens. Count before writing.
Every project/concept name links to its article if one exists.
Sort by relevance (source count, star count) not alphabetically, except where noted.
Missing.md excludes entities that already have reference cards.