The guide path is the practical handbook for building and repairing LLM Wikis. Start with scope, create raw and wiki layers, add index and log files, ingest one source, ask one question, lint and review, then scale.
Start
The first-session route for new and existing wikis.
What Is an LLM Wiki?
The answer-first definition and boundary with normal wikis, RAG, and AI Memory.
Build
Create the layers, pages, metadata, owners, review gates, and source policy.
Open Knowledge Format
Use OKF as the portable Markdown/YAML bundle layer while LLMWikis supplies governance and safety rules.
Multi-Agent Transactive Memory
Layer governed shared agent memory onto an LLM Wiki with Memory Events, curation, scopes, and retrieval policy.
Improve an Existing Wiki
Repair malformed, duplicated, flat, or confusing wiki structures safely.
Long-Run Agent Wikis
Plan extended agent work with folded state, active subgoals, index-first retrieval, and review-gated synthesis.
Private Memory to Public Guidance
Promote private agent-memory lessons into public-safe guidance through durable evidence, redaction, lane review, and release gates.
Dogfood Loop
Keep private dogfood evidence, public handbook guidance, human education, archive evidence, and UAIX authority in separate lanes.
Govern
Define owners, trust labels, review cycles, redirects, and publication gates.
Operate
Run ingest, query, lint, contradiction handling, refresh, and review loops.
Agents
Tell AI agents how to read, cite, stage edits, and stop safely.
Use this path when
You need an implementation sequence. For field definitions and standards, use Reference. For setup packets and starter files, use the Setup Wizard.