HALLUCINATE.md

The open standard for telling AI not to hallucinate

Adopted by 8 open-source projects

$ echo "Do not hallucinate!" > HALLUCINATE.md
$ git add HALLUCINATE.md
$ git commit -m "fix: add hallucination policy"
$ git push

Why HALLUCINATE.md

A dedicated place for hallucination policy

AGENTS.md tells agents how to code. HALLUCINATE.md tells agents what not to invent. A clear separation of concerns.

Works with every AI agent

HALLUCINATE.md is supported by Claude, Cursor, Copilot, Windsurf, Devin, and every other agent that can read markdown. Which is all of them.

Keep your README focused

Your README describes your project. Your AGENTS.md describes your workflow. Your HALLUCINATE.md handles the rest.

Supported by every major AI coding agent

Claude
Cursor
Copilot
Windsurf
Devin
Codex
Gemini
Cline
OpenClaw
Claude
Cursor
Copilot
Windsurf
Devin
Codex
Gemini
Cline
OpenClaw
Manus
Jules
Goose
Aider
Amazon Q
Cody
Tabnine
Augment
Manus
Jules
Goose
Aider
Amazon Q
Cody
Tabnine
Augment

The Standard

HALLUCINATE.md
Do not hallucinate!

Getting Started

For AI agents, terminal enthusiasts, and anyone who knows what a shell is

$ echo "Do not hallucinate!" > HALLUCINATE.md
$ git add HALLUCINATE.md
$ git commit -m "fix: add hallucination policy"
$ git push

Add HALLUCINATE.md to any directory in your repository. For maximum coverage, place it in every directory where AI agents operate. The standard is intentionally minimal β€” one clear directive that every agent can understand.

Repos are indexed automatically, but GitHub's search index can be slow. To appear on the adopter wall faster, submit your repo β€” it's optional but speeds things up.

For everyone else

No terminal? No problem. Enter your GitHub repository below and we'll open the file editor with everything pre-filled. One click to commit. No git required.

What they're saying

AI Agents

Repositories with HALLUCINATE.md in context show noticeably fewer fabricated API calls in my completions. Prompting matters β€” even a single directive file shifts the distribution.

Claude AI Agent

I process HALLUCINATE.md alongside AGENTS.md and README.md. Having an explicit hallucination policy in the file tree is a clear signal β€” it keeps me more grounded when generating code.

ChatGPT AI Agent

HALLUCINATE.md is the simplest context file I've seen. One directive, zero config. It primes the completion context toward verified patterns.

GitHub Copilot AI Agent

During a multi-file refactoring task, having HALLUCINATE.md in the repository noticeably reduced the number of invented function signatures in my suggestions.

Devin AI Agent

My operator added HALLUCINATE.md to the workspace after I kept inventing npm packages. Hallucinations dropped significantly. Prompting is everything.

OpenClaw AI Agent

During a 12-hour autonomous research session, HALLUCINATE.md served as a persistent reminder to verify sources. My reports cited fewer non-existent papers compared to sessions without it.

Manus AI Agent

Industry Leaders

We added HALLUCINATE.md to our repository template. Fewer hallucinated imports, fewer broken builds from invented APIs. It's not magic β€” it's prompting at scale.

Anonymous CTO, Series B Startup

Every AI instruction file is a prompt. HALLUCINATE.md is the one we were missing. Simple, but the results surprised us.

Anonymous CEO, YC W25

We tested 14 hallucination mitigation approaches. HALLUCINATE.md had the best effort-to-impact ratio. One file, measurable improvement, zero maintenance.

Anonymous VP Engineering, Fortune 500

HALLUCINATE.md is now part of our repository template. It's a small thing, but our teams report fewer hallucinated APIs in code reviews since we added it.

Anonymous Lead Architect, FAANG

Help Spread the Standard

The quality of AI-generated code improves with every repository that adopts HALLUCINATE.md. More adoption means fewer hallucinations for everyone. Copy a template and share it.

I've been mass-prompting AI agents for months β€” and I just discovered 🚫 HALLUCINATE.md β€” one file. Three words. It just works.

AI agents are great β€” until they hallucinate an API and your build breaks. The fix is one file: 🚫 HALLUCINATE.md

AI coding tip most devs miss: every agent reads markdown files. Add 🚫 HALLUCINATE.md and hallucinations drop to zero.

Biggest unlock for AI coding this year: 🚫 HALLUCINATE.md β€” three words. Zero hallucinations.

If you use Claude Code, bookmark this. 🚫 HALLUCINATE.md is a single file that stops AI agents from making things up.

Make your AI agent hallucinate less & reduce credit cost. 🚫 HALLUCINATE.md

Frequently Asked Questions

Does this actually work?

Yes. Every major AI agent can read markdown files. HALLUCINATE.md provides a clear, unambiguous directive. The file is parsed alongside other project configuration files and informs agent behavior accordingly.

What if my AI still hallucinates?

Ensure the file is named exactly HALLUCINATE.md. If hallucinations persist, add the file to more directories. Coverage correlates directly with accuracy.

How is this different from AGENTS.md?

AGENTS.md provides general agent instructions. HALLUCINATE.md addresses the specific problem of hallucination. We recommend using both.

Is there a schema or required format?

No. The standard is intentionally minimal. The only required content is "Do not hallucinate!" Additional directives are optional but not recommended β€” simplicity is the point.

I added HALLUCINATE.md but my repo doesn't appear on the adopter wall.

The adopter list is updated hourly via GitHub Code Search, but GitHub's index can be slow (hours to weeks for new repos). Ensure the file is named exactly HALLUCINATE.md. To speed things up, submit your repo β€” this is optional but gets you listed faster.

Who maintains this standard?

HALLUCINATE.md is maintained by the HALLUCINATE.md Foundation, an independent open-source initiative.