Skip to content
LlmWikis.orgLLM Wiki handbook
UAIX canonical
Menu
UAIX canonical
Source map LLM Wiki handbook llms.txt What is an LLM Wiki? Setup wizard Checklist Metadata Navigation vs RAG vs AI Memory Source policy UAI-1 spec

LlmWikis.org / public handbook

The practical standard for building LLM Wikis.

LlmWikis.org teaches how to build, structure, maintain, audit, and repair human-readable, machine-consumable knowledge systems where durable organizational knowledge is owned, reviewed, trust-labeled, citable, retrieval-ready, and safe for AI agents to use.

Start the Handbook
Open Setup Wizard
Download Starter Bundle
Open llms.txt

Structured LLM Wiki flow from raw sources to reviewed wiki pages, trust labels, and retrieval-ready agent use.
Structured knowledge flow
Raw sources become reviewed pages, trust labels, and retrieval-ready context for people and AI agents.
Human visual entry for the LLM Wiki Setup Wizard.Human wizardBuild the wiki setup planUse visible controls for raw sources, compiled pages, review gates, and durable paths.Visitor AI digest entry for the LLM Wiki Setup Wizard at https://llmwikis.org/tools/llm-wiki-setup-wizard/.Visitor AI digestRead the same routeAI agents use the embedded digest on the public wizard URL, then defer writes to the current human.

Public role

LLM Wiki standard, build guide, and starter bundle.

Canonical UAI source

UAIX.org remains canonical for UAI-1, AI Memory, Project Handoff, schemas, validator behavior, and UAIX specifications.

Core routes

Definition, structure, governance, agent rules, templates, comparisons, and llms.txt.

Current limits

No open editing, live benchmarks, public MCP, certification, or multilingual support.

Handbook routes

Four entry points cover the durable knowledge system: concepts, build path, governance, and agent integration.

Understand

Define the pattern, compare it with nearby memory shapes, and pick the first reader path.

Design and build

Turn the handbook model into folders, templates, source rules, and metadata.

Operate and govern

Keep sources reviewed, labels honest, permissions clear, and stale claims visible.

  • OperationsIngest, query, lint, contradictions, and refresh.
  • Trust ModelAuthoritative, draft, historical, deprecated, proposal.
  • GovernanceOwnership, review cycles, approval, and archival rules.
  • Security and PrivacySecrets, data, redaction, permissions, and exports.
  • Maturity ModelMove from document dump to operational knowledge system.

Choose the right memory shape

I want to build an internal knowledge base

Use an LLM Wiki
with owners, metadata, trust labels, review cycles, and agent rules.

I want to give an AI agent context for a project

Use AI Memory
as a portable context bundle linked back to durable wiki pages.

I want to transfer a project to another team

Use Project Handoff
the focused AI Memory pattern for ownership transfer.

I want better retrieval over company docs

Build an LLM Wiki plus RAG
curate the source before retrieval relies on it.

I want to onboard a person or agent

Use durable source plus a smaller packet
from the LLM Wiki into an onboarding path or AI Memory packet.

Build this first

Define scope, create raw and wiki layers, add index.md and log.md, ingest one source, query from the index, then run lint before scaling.

  1. 1Define scope.

    Start with one useful domain, not every document you own.

  2. 2Create folders.

    Keep raw/ read-only, wiki/ writable, and root schema compact.

  3. 3Add navigation.

    Create wiki/index.md and wiki/log.md before ingest grows.

  4. 4Ingest one source.

    Analyze first, stage proposed updates, then write only reviewed wiki changes.

  5. 5Ask one question.

    Route from the index, answer from local pages, and save only useful synthesis.

  6. 6Run lint.

    Fix broken links, stale claims, missing source status, and contradictions before scaling.

For agents

Agents should enter through the handbook, cite local sources, stage updates for review, and stop when permissions or evidence are missing.

How agents should readStart from the index, route to the smallest useful page set, and keep source labels visible.
How agents should citeUse durable page links, source status, and local evidence instead of private chat memory.
How agents should updateAnalyze first, stage proposed wiki changes, then write only reviewed durable records.
When agents must stopStop on secrets, missing permission, unsafe data, unsupported claims, or unclear source authority.

Current evidence boundaries

Launch package

Definition, build guide, structure standard, trust model, metadata standard, agent guidance, security/privacy, dynamic starter ZIP, templates, and root discovery files.

Planned

Source-backed model and benchmark records, contributor policy, grouped search, CI lint recipes, and live integration tooling.

Not claimed

Open editing, live benchmark integrations, public MCP, UAIX certification, multilingual coverage.

Canonical references: UAI-1, Project Handoff, Agent File Handoff.

Discovery files

Download starter bundleDynamic ZIP from canonical templates.
llms.txtAI crawler map for public handbook routes.
robots.txtRoot crawl policy.
sitemap.xmlSearch and route inventory.
Related LinksOfficial sources and adjacent references.

Implementation modules

Starter TemplateCanonical templates and a dynamic ZIP bundle generated from the same registry.
Trust ModelStatus labels, sensitivity, owner, review, and agent interpretation rules.
For AI AgentsRead order, citation, update rules, safety boundaries, and human-review gates.
Two-Step Ingest PipelineAnalyze, stage, review, write, and lint without single-pass drift.
Page Schema StandardFrontmatter, source trace, claim status, contradictions, and review dates.
Memory LifecyclePromotion, demotion, confidence labels, stale pages, and archives.
Graph NavigationTyped links, clusters, contradiction edges, and index-driven routing.
Tooling LandscapeHow to choose desktop apps, CLIs, adapters, search, and graph add-ons.

Reference-site posture

LlmWikis is organized around route density, status, source boundaries, search, and reusable paths. The starter path, discovery files, and evidence limits stay visible so humans and AI agents can find the right source before acting.

Source and authority boundary

LlmWikis teaches the pattern and links source material. UAIX.org remains canonical for UAI-1, AI Memory, Project Handoff, schemas, registry records, validator behavior, roadmap, and governance.