These are three of our five production-grade schema.md templates for the LLM Wiki pattern, free to copy below. Drop your email if you want the full pack (5 schemas + ingest scripts + a 15-minute video walkthrough) sent to your inbox the moment it ships. Brand new to the pattern? Read the LLM Wiki primer first — these schemas make a lot more sense once you know what they are doing.
One email at launch, plus the occasional schema-pattern digest. Unsubscribe in one click.
The three schemas on this page cover the most common LLM Wiki use cases. Each one is the result of dozens of real wiki compilations — they are tuned to produce clean, predictable output from Claude, Gemini, or any capable LLM.
Two more schemas (Product and SEO) are in the Starter Kit along with the ingest pipeline and the video walkthrough. The free schemas below cover ~70% of personal use cases on their own.
The everyday default. Use this for tracking ideas, books, articles, and people across any domain. It is the schema we recommend everyone starts with.
# Schema: Personal LLM Wiki
## Entity types
- **concept**: an idea, technique, or abstraction
- **person**: someone whose work I reference
- **tool**: software, library, or service I use
- **source**: a paper, article, video, or book I read
## Page format
Every wiki page starts with frontmatter:
type: concept | person | tool | source
tags: [list]
created: YYYY-MM-DD
updated: YYYY-MM-DD
Then the body in this order:
1. Summary (1-2 sentences, what this is)
2. Key points (3-5 bullets)
3. Related (wikilinks to other pages)
4. Open questions (things I have not yet resolved)
5. Source refs (links to raw/ files)
## Rules
- Use [[wikilinks]] for anything that is itself a page
- Never overwrite Open Questions without flagging
- When two sources contradict, note it in Open Questions
- Keep each page under 500 words
- Mark speculation clearly with "Hypothesis:" prefixBest for: writers, consultants, anyone tracking a single domain across many sources. Pairs nicely with the Obsidian setup guide.
For grad students, post-docs, analysts, and anyone managing 50+ academic papers. This schema treats citations and methodology as first-class entities so your wiki captures the structure of a literature review automatically.
# Schema: Research LLM Wiki
## Entity types
- **paper**: a publication (title, authors, year, venue, methods, findings)
- **author**: a researcher you track (affiliation, key papers)
- **concept**: a theoretical or methodological idea
- **dataset**: a named dataset (what it measures, how collected)
- **method**: a named methodology or technique
## Relationships (use wikilinks)
- paper CITES paper
- paper EXTENDS paper
- paper CONTRADICTS paper
- paper USES method
- paper USES dataset
- author AUTHORED paper
- concept DEFINED-IN paper
- concept DEBATED-IN paper
## Page format
Frontmatter:
type: paper | author | concept | dataset | method
year: YYYY
topic: [topic tags]
status: read | skimmed | queued
Body:
1. One-sentence summary
2. Key claim or contribution
3. Method / evidence
4. Limitations
5. Links (wikilinks)
6. Open questions
## Rules
- Every paper page must cite at least one other page
- Contradictions go in Open Questions, never silently overwritten
- Keep each page under 600 words
- Mark speculation with "Hypothesis:" prefixBest for: literature reviews, dissertation work, multi-paper synthesis. The LLM Wiki for research walkthrough shows how to wire this to Zotero.
For developers building any non-trivial project. Use this in your project's wiki/ directory alongside the code, fed by Claude Code or Cursor.
# Schema: Engineering LLM Wiki
## Entity types
- **module**: a code subsystem with clear boundaries
- **decision**: an architectural or design decision (ADR-style)
- **issue**: a known bug, gotcha, or open problem
- **dependency**: an external library or service we depend on
- **runbook**: an operational procedure (deploy, debug, restore)
## Page format
Frontmatter:
type: module | decision | issue | dependency | runbook
status: active | deprecated | resolved
owner: [name or team]
updated: YYYY-MM-DD
Body for module:
1. Purpose (1 sentence)
2. Public API (the surface other modules see)
3. Internal architecture (3-6 bullets)
4. Dependencies (wikilinks)
5. Decisions (wikilinks to ADRs)
6. Known issues (wikilinks)
Body for decision:
1. Context (what we needed to decide)
2. Options considered
3. Decision and rationale
4. Tradeoffs accepted
5. Date and status
## Rules
- Every module must list its public API
- Every decision must have a date
- Issues stay on the page until status: resolved
- Deprecations are never deleted, only marked
- Keep each page under 800 words (decisions can be longer)Best for: engineering teams, indie developers, anyone using Claude Code on a multi-month project. Pairs with the Claude Code tutorial.
schema.md in the root of your wiki folder (or in your Obsidian vault's meta/ folder)raw/ — a PDF, an article, a meeting note, anythingwiki/ folder, then read it like you have never seen the sourceThe first three compiles are about tuning the schema to your style. After that, the wiki starts compounding — new sources update existing pages instead of creating orphans.
The two remaining schemas (Product and SEO) plus the ingest.py and lint.py scripts and a 15-minute video walkthrough live in the LLM Wiki Starter Kit. It is one purchase, lifetime updates, $19 launch price.
If you are happy with the three free schemas above, you do not need to buy anything — that is genuinely the point of giving them away.