Wiki Schema
Ground rules for the agent-curated knowledge base. This wiki is agent-first: optimized for LLM-driven search, ingestion, and synthesis. The human-optimized content lives in the Digital Garden (sibling sections in this vault).
Domain
General purpose — research, reference, and synthesis across any topic David explores. Covers entities (people, orgs, products), concepts (ideas, techniques, phenomena), and comparisons (side-by-side analyses).
Conventions
- File names: lowercase, hyphens, no spaces (e.g.,
transformer-architecture.md) - Every wiki page starts with YAML frontmatter (see below)
- Use
[[wikilinks]]to link between pages — minimum 2 outbound links per page - Update
updateddate when modifying a page - Every new page must be added to
index.mdunder the correct section - Every action appended to
log.md
Frontmatter
---
title: Page Title
created: YYYY-MM-DD
updated: YYYY-MM-DD
type: entity | concept | comparison | query | summary
tags: [domain/topic]
sources: [raw/articles/source-name.md]
confidence: high | medium | low # how well-supported the claims are
contested: true # set when page has unresolved contradictions
contradictions: [other-page-slug] # pages this one conflicts with
---confidence and contested are optional but recommended for opinion-heavy topics.
Raw Source Frontmatter
Immutable source material also gets small frontmatter for drift detection:
---
source_url: https://example.com/article
ingested: YYYY-MM-DD
sha256: <hex digest of body below frontmatter>
---Tag Taxonomy
- AI/ML: model, architecture, benchmark, training, inference, alignment
- Tools: tool, software, framework, cli, api, protocol
- People/Orgs: person, company, research-lab, open-source
- Health/Fitness: exercise, nutrition, recovery, ergonomics
- Family: parenting, child-development, education, activities
- Finance: investing, budgeting, insurance, tax
- Home: renovation, appliance, repair, maintenance
- Meta: comparison, timeline, controversy, prediction, reference
Add new tags here BEFORE using them.
Page Thresholds
- Create a page when an entity/concept appears in 2+ sources OR is central to one source
- Add to existing page when a source mentions something already covered
- DON’T create a page for passing mentions or minor details
- Split a page when it exceeds ~200 lines — break into sub-topics
Quality Signals (cross-pollinated to Digital Garden)
These patterns apply to both the wiki and the garden’s research/shopping notes:
- Provenance markers: On pages synthesizing 3+ sources, append
^[raw/articles/source.md]at paragraph end to trace claims back to source - Confidence: Mark
confidence: lowfor single-source or speculative claims. Don’t let weak claims harden into accepted fact - Contested: When sources disagree, note both positions with dates rather than silently overwriting
- Source drift: On re-ingest, recompute sha256 of raw sources — flag mismatches
Layer Structure
wiki/
├── SCHEMA.md # This file
├── index.md # Content catalog
├── log.md # Action log (append-only)
├── raw/ # Layer 1: Immutable source material
│ ├── articles/ # Web articles, clippings
│ ├── papers/ # PDFs, arxiv papers
│ ├── transcripts/ # Meeting notes, interviews
│ └── assets/ # Images, diagrams
├── entities/ # Layer 2: People, orgs, products, models
├── concepts/ # Layer 2: Ideas, techniques, topics
├── comparisons/ # Layer 2: Side-by-side analyses
└── queries/ # Layer 2: Filed query results worth keeping