What it finds on a demo project

The CLI reads AGENTS.md, CLAUDE.md and the other instruction files coding tools use, and flags near-empty ones, files that never mention each other and values that resemble credentials.

SiteCMD CLI 2.0.0
Demo project: Agent instruction files
npx sitecmd audit .
SiteCMD Code Scan - .
Checked: November 9, 2026
Framework: Express
Issues: 3 total | critical 0 | high 1 | medium 0 | low 2

High (1)
- [AI Setup · ai-scaffolding] Credential-shaped literal in CLAUDE.md - CLAUDE.md:5

Low (2)
- [AI Setup · ai-scaffolding] AGENTS.md has almost no guidance - AGENTS.md:1
- [AI Setup · ai-scaffolding] AI instruction files can drift out of sync - CLAUDE.md:1

Your code passes? Now watch your site.

Join the waitlist

Follow AGENTS.md best practices for commands, tests and style

Include the build and test commands, code style and testing steps an agent needs. These are the sections the AGENTS.md format suggests. In a monorepo, a nested AGENTS.md in each package overrides the root one for files there.

Connect AGENTS.md and CLAUDE.md to one source of guidance

Keep shared instructions in AGENTS.md and point the other files to it. Claude Code reads CLAUDE.md instead of AGENTS.md, so if your project has both, CLAUDE.md can import AGENTS.md. Claude Code's docs explain how.

Keep credentials out of CLAUDE.md and agent configuration

Explain where a tool gets a credential without putting the credential itself in the instructions. Review a flagged value without copying it into a new issue or log. A placeholder or public identifier can match too.

Frequently asked questions

What's the difference between AGENTS.md and CLAUDE.md?

AGENTS.md is an open format shared across coding tools; CLAUDE.md is specific to Claude Code.

Confirm you're a person