agentsclimarketplace

Llmwiki init

Skill sergsrgsg/rock-star-skills/skills/llmwiki-init

Initialize the LLM Wiki. Creates directory structure, index, log, sets up qmd collection, and runs initial ingest of all existing content from raw/, docs/, and notes/. Run once per project.From its SKILL.md

Install
npx -y skills add sergsrgsg/rock-star-skills --skill llmwiki-init

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

One thing to look at

  • 0 stars0 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.

SKILL.md

4.3 KB, 955 tokens by cl100k_base, as published. Nobody here has run it

LLM Wiki: Initialize

Set up a new LLM Wiki in the current project and run the first full ingest.

Pre-flight

  1. Check if wiki/index.md already exists. If it does and has content beyond the template, ask the user whether to reinitialize (destructive) or skip.
  2. Verify raw/, docs/, and notes/ directories exist. Create any that are missing.
  3. Verify raw/attachments/ exists. Create if missing.

Step 1: Create Wiki Structure

Ensure the following directories and files exist:

wiki/
├── index.md          # Master index of all wiki pages
├── log.md            # Chronological log of operations
├── entities/         # Pages about specific things (people, companies, products, tools)
├── concepts/         # Conceptual/topic pages
├── sources/          # Source summaries (one per ingested source)
├── comparisons/      # Comparison tables, side-by-side analyses
└── synthesis/        # Cross-cutting analyses, insights

If files already exist, leave them as-is.

Step 2: Set Up qmd Collection

Run:

qmd collection add wiki ./wiki

This registers the wiki/ directory as a qmd collection named "wiki". If the collection already exists, skip.

Then update the index:

qmd update

If qmd is not installed, warn the user: npm install -g @tobilu/qmd and continue without it.

Step 3: Initial Ingest

Collect ALL existing files that should be ingested:

  1. raw/ — all markdown files and other documents (skip raw/attachments/)
  2. docs/ — all markdown files recursively
  3. notes/ — all markdown files recursively

For each file found, invoke the llmwiki:ingest skill's logic (or invoke the skill itself). Process files in batches using parallel subagents where possible:

  • Group files by directory/topic
  • For each batch, spawn a subagent to read the files and produce wiki pages
  • Each subagent should:
    • Read the source file
    • Determine what wiki pages to create/update (entity pages, concept pages, source summary)
    • Write the source summary to wiki/sources/
    • Create or update entity/concept pages in wiki/entities/ and wiki/concepts/
    • Add [[wikilinks]] between related pages
    • Return the list of pages created/updated

After all subagents complete:

  • Rebuild wiki/index.md with all pages
  • Append a log entry to wiki/log.md:
    ## [YYYY-MM-DD] init | Initialized wiki with N sources, M pages created
    

Step 4: Build Embeddings (Optional)

If qmd is available, run:

qmd embed

This builds vector embeddings for semantic search. If it fails or takes too long, skip — BM25 search still works.

Step 5: Report

Print a summary:

  • Number of source files ingested
  • Number of wiki pages created (by category)
  • Any files that failed or were skipped
  • qmd status (collection registered, embedding status)

Wiki Page Conventions

All wiki pages MUST follow these conventions:

Frontmatter

Every wiki page has YAML frontmatter:

---
title: Page Title
type: entity | concept | source | comparison | synthesis
tags: [tag1, tag2]
sources: [relative/path/to/source.md]
created: YYYY-MM-DD
updated: YYYY-MM-DD
---

Wikilinks

Use [[Page Title]] syntax for cross-references. Obsidian resolves these automatically. When creating a new page, add wikilinks to all related existing pages, and update those existing pages to link back.

Source Summaries

Source summary filenames: kebab-case version of the source title, in wiki/sources/. Example: wiki/sources/cross-platform-tech-stacks-2026.md.

Entity & Concept Pages

Use descriptive kebab-case filenames. Example: wiki/entities/expo.md, wiki/concepts/local-first-architecture.md.

File Naming

  • All wiki page filenames: lowercase kebab-case with .md extension
  • No spaces, no special characters beyond hyphens
  • Keep names concise but descriptive

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 326,970. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.