About This Wiki|7 min read|Last reviewed 2026-04-22

Using This Wiki With AI

The JB wiki can be useful inside an AI workflow, but only if the model preserves the parts of the site that keep the claims honest. If a system ingests the topical pages and drops the evidence tags, the verification statuses, or the species boundaries, it is no longer representing the wiki accurately.

This page exists to make responsible AI use explicit rather than assumed.

What This Page Explains

This page explains how to use the JB wiki with ChatGPT, Claude, NotebookLM, a custom retrieval stack, or another frontier model without flattening the methodology that governs the content.

It also gives the reading order that should come before broader ingestion, names the role of /llms.txt, and states the reuse posture for this knowledge-base layer.

Core Explanation

The first thing to understand is that the wiki is not just a pile of topical pages. It has a visible claim-discipline layer. Evidence tags tell you how strongly a claim may be stated. Verification status tells you how much of the citation chain has been walked. The SCR references and sources show where those claims are coming from. Any AI workflow that strips those layers out is simplifying away part of the actual meaning.

The second thing to understand is that JB uses some vocabulary in a distinctive way. A model that ingests pillar pages without the crosswalk page will often translate JB language into the wrong mainstream equivalent and then answer confidently from the mistranslation.

The third thing to understand is that cross-species claims need two-step handling. "Documented in another species" is not the same as "documented in dogs." If your AI notes or outputs collapse that distinction, you are overstating the wiki.

How to Use This

Ingestion Order Protocol

Read What This Wiki Is and Isn't so the system's scope and limits are explicit before topical ingestion begins.
Read How to Read an Entry so the model understands the structure of a normal wiki page.
Read Evidence Tags so the model preserves rhetorical ceilings instead of treating every page as equally settled.
Read Verification so the model keeps citation-integrity status separate from evidence strength.
Read Scientific Claims Register so the model understands where the evidence discipline is coming from.
Read Five Pillars Terminological Crosswalk so JB language is translated carefully rather than flattened.
Only then ingest topical categories or individual pages relevant to the user's question.

Operating Rules for AI Use

Preserve the evidence tag when summarizing a claim.
Preserve verification status separately from the evidence tag.
Name the studied species whenever the support is not canine-direct.
Do not rewrite heuristic claims as settled scientific conclusions.
Prefer citing the exact wiki page you used rather than speaking in generalities about "the JB method."

For machine-readable orientation, JB publishes /llms.txt. That file is the shortest honest map of the wiki for LLM and retrieval use. It points to the orientation pages first and only then to the topical categories.

If you are building a custom assistant, the safest prompt frame is something like this:

Use the Just Behaving wiki as a tagged knowledge base. Preserve evidence tags as rhetorical ceilings, keep verification status separate from evidence level, preserve species boundaries, and cite the specific JB wiki page used for any answer.

Citation and Reuse

Public guidance in this knowledge-base layer is offered for non-commercial reuse with attribution under CC BY-NC 4.0.

That means research, educational, and other non-commercial reuse is allowed with attribution. It does not mean unrestricted commercial repackaging. If you quote or summarize JB content in an AI workflow, cite the specific page used.

Sources or Governing References

Public machine-readable entry point: /llms.txt
Internal authority: HWW, the Wiki Specification, and the Scientific Claims Register governance model
Public companion pages in this category: What This Wiki Is and Isn't, How to Read an Entry, Evidence Tags, Verification, and the Terminological Crosswalk