geronimo-iia/llm-wiki-skills/frontmatter

Claude Code and agent skill by geronimo-iia/llm-wiki-skills

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Reference for writing correct wiki frontmatter — required fields, field conventions, type-specific validation, and the accumulation contract.

Install

npx vskill@latest install geronimo-iia/llm-wiki-skills/frontmatter

SKILL.md

Frontmatter Reference

Background knowledge for writing correct frontmatter in wiki pages. This skill is loaded automatically when writing wiki pages.

For type system management (listing, adding, removing types), see the schema skill. For the full type taxonomy, see references/type-taxonomy.md.

Quick template

Get the exact frontmatter scaffold for any type:

wiki_schema(action: "show", type: "<type>", template: true)

Required fields

Every wiki page must have:

FieldRule
titleConcise, specific, unambiguous
typePage type from the type registry

Additional required fields depend on the type:

TypeExtra required fields
concept, query-resultread_when
skillname, description (instead of title, summary)

Field conventions

title
TypeConventionExample
conceptThe concept name"Mixture of Experts"
paper"Paper Title (Year)""Switch Transformer (2021)"
article"Article Title""Why MoE Models Are the Future"
documentation"Product — Topic""vLLM — MoE Configuration"
transcript"Event — Topic (Date)""Team Standup — MoE Decision (2025-07-15)"
noteDescriptive title"Quick Notes on Routing Strategies"
clippingOriginal title or descriptive"Karpathy on LLM Wikis"
book-chapter"Book Title — Chapter""Designing ML Systems — Ch. 8"
thread"Forum — Topic""HN — MoE Scaling Discussion"
dataDescriptive title"MoE Benchmark Results 2024"
query-result"Topic — Aspect""MoE Routing — Design Decision"
sectionThe section name"Scaling Research"

Do not include the page type in the title ("Paper: Switch Transformer" ✗).

summary vs tldr
  • summary describes scope ("what is this page about?")
  • tldr states the conclusion ("what's the bottom line?")
read_when

Write as situations, not keywords:

yaml
read_when:
- "Reasoning about MoE architecture tradeoffs"
- "Comparing sparse vs dense model scaling"

Not: ["MoE", "scaling", "routing"] — that's what tags is for.

tags

Flat list, lowercase, hyphenated:

yaml
tags: [mixture-of-experts, scaling, transformers]
aliases

Alternative names and synonyms for this concept — keyword-indexed, so synonym queries in wiki_search hit this field directly:

yaml
aliases: [Mixture-of-Experts, MoE, sparse routing]

Use for acronyms, prior names, and common informal variants. Lowercase preferred but not required — BM25 keyword matching is exact on FAST fields.

sources

Slugs of source pages that actually contributed claims:

yaml
sources: [sources/switch-transformer-2021]
concepts

Slugs of concept pages this page directly discusses or depends on:

yaml
concepts: [concepts/scaling-laws]

Apply the backlink quality test: would a reader benefit from navigating there?

confidence

A float 0.0–1.0 on both page and claim objects. Same scale everywhere.

ValueWhen to use
0.9Multiple corroborating sources, well-established
0.5Single source, or source has caveats
0.2Preliminary, speculative, or contradicted

Default to 0.5 when unsure. Omit on claim objects when no certainty signal was recorded (absence is distinct from 0.5 neutral).

claims

Structured claims extracted from sources:

yaml
claims:
- text: "Sparse MoE reduces effective compute 8x"
confidence: 0.9
source: sources/switch-transformer-2021
section: "Results"
owner

Who is responsible — a person, team, or agent session ID.

superseded_by

Slug of the page that replaces this one. Use when a page is effectively replaced, not for minor updates. Creates a superseded-by graph edge.

Graph edges from frontmatter fields

Certain frontmatter fields create typed graph edges at ingest:

TypeFieldRelationTarget types
conceptsourcesfed-byAll source types
conceptconceptsdepends-onconcept
source typessourcescitesAll source types
source typesconceptsinformsconcept
docsourcesinformed-byAll source types
skilldocument_refsdocumented-bydoc
anysuperseded_bysuperseded-byAny

Body [[wiki-links]] get a generic links-to relation.

Field aliasing

Skill pages use different field names. The engine aliases them at ingest time so they index uniformly:

Source fieldCanonical field
nametitle
descriptionsummary
when_to_useread_when

Aliases affect indexing only — files on disk are never rewritten.

Accumulation contract

When updating a page:

  1. Read the current page first — wiki_content_read
  2. Preserve existing list values — do not drop tags, read_when,

sources, concepts, or claims

  1. Add new values to lists, do not replace them
  2. Update scalar fields only with clear reason
  3. Write the complete file, then wiki_ingest

Common mistakes

MistakeFix
Missing titleEngine rejects the file
Using source-summary as typePick specific source type
Missing read_whenAlways include 2–5 retrieval conditions
Listing every related sourceOnly sources that contributed claims
Dropping existing tags on updateRead first, preserve existing values
Setting confidence: 0.9 without evidenceDefault to 0.5
Classifying by topic instead of source natureBlog about research → article

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/geronimo-iia/llm-wiki-skills/frontmatter/badge)](https://verified-skill.com/skills/geronimo-iia/llm-wiki-skills/frontmatter)

Something wrong with this skill or its page? Tell us.