← Back to Terminology Glossarytools

AGENTS.md

An open standard repository configuration file that provides operating guidelines, architecture maps, and verification rules for autonomous AI coding agents.

What is AGENTS.md?

AGENTS.md is an open-standard context configuration file placed at the root of a code repository.

As autonomous AI agents (Claude Code, Cursor, Antigravity, Devin, Copilot Workspace) enter a project, they inspect AGENTS.md to immediately understand the repository's coding standards, folder hierarchy, forbidden dependencies, and verification commands.

---

Copyable Enterprise AGENTS.md Blueprint

markdown
# AGENTS.md β€” Repository Guidelines for AI Agents

## 1. Core Architecture Principles
- Framework: Next.js 14+ App Router.
- Typing: Strict TypeScript (`tsc --noEmit` must pass).
- Styling: Vanilla CSS Modules with custom properties. No inline styles.

## 2. Directory Structure
- `app/`: Pages, layouts, and API route handlers.
- `app/components/`: Shared, atomic UI components.
- `app/lib/`: Business logic, Firestore/Postgres helpers.

## 3. Mandatory Error Conventions
- All server routes MUST return `{ success: boolean, data?: any, error?: string }`.
- Log full error tracebacks silently; never leak stack traces to client viewports.

## 4. Verification Protocol
Run `npm run build` and `npm test` before requesting human review.

---

References

For additional agent configuration standards, visit lopopolo/harness-engineering.