AGENTS.md

Repository-wide AGENTS.md baseline

A concise project contract covering setup, validation, architecture, safety, and commits.

Revision: 1
Verified: 2026-07-26

Save or explore

#agents #repository #instructions

Compatibility and paths

CodexAGENTS.mdRepository and nested project folders

GitHub CopilotAGENTS.mdRepository and nested project folders

VS CodeAGENTS.mdRepository and nested project folders

Trust and provenance

Curated record reviewed 2026-07-26. Results still depend on the supplied context and target environment.

Fill variables

Values stay in this browser tab and are not stored.

Install CommandrequiredValidation CommandrequiredTest CommandrequiredDev CommandrequiredArchitecture Rulerequired

Generated asset5 required field(s) must be completed

# Repository-wide AGENTS.md baseline

## Scope and precedence
- Apply these rules from this directory through its descendants.
- Follow a more specific nested `AGENTS.md` for files in that subtree.
- User and system instructions override this file.

## Repository setup
- Install dependencies with `{{installCommand}}`.
- Use the repository-pinned runtime and package manager.

## Validation and tests
- Run `{{validationCommand}}` after focused changes.
- Run `{{testCommand}}` before handoff.
- Report commands that could not run and the exact reason.

## Task-specific rules
# Project instructions

## Setup
- Install dependencies with `{{installCommand}}`.
- Start development with `{{devCommand}}`.

## Validation
- Run `{{testCommand}}` before handoff.

## Architecture
- Preserve {{architectureRule}}.

## Safety
- Never commit secrets or generated credentials.
- Confirm destructive operations before execution.

## Version control
- Commit each logical change separately.

## Safety and protected paths
- Never expose secrets, credentials, private keys, or production data.
- Do not edit generated files, lockfiles, migrations, or deployment state unless the task explicitly requires it.
- Ask before destructive commands, irreversible migrations, privilege changes, or external publication.

## Completion
- Keep the change bounded to the requested scope.
- Summarize modified files, validation evidence, remaining risks, and any manual follow-up.

Real example

Input

Project uses AGENTS.md with the required variables filled.

Expected result

# Repository-wide AGENTS.md baseline

## Scope and precedence
- Apply these rules from this directory through its descendants.
- Follow a more specific nested `AGENTS.md` for files in that subtree.
- User and system instructions override this file.

## Repository setup
- Install dependencies with `example-value`.
- Use the repository-pinned runtime and package manager.

## Validation and tests
- Run `example-value` after focused changes.
- Run `example-value` before handoff.
- Report commands that could not run and the exact reason.

## Task-specific rules
# Project instructions

## Setup
- Install dependencies with `example-value`.
- Start development with `example-value`.

## Validation
- Run `example-value` before handoff.

## Architecture
- Preserve example-value.

## Safety
- Never commit secrets or generated credentials.
- Confirm destructive operations before execution.

## Version control
- Commit each logical change separately.

## Safety and protected paths
- Never expose secrets, credentials, private keys, or production data.
- Do not edit generated files, lockfiles, migrations, or deployment state unless the task explicitly requires it.
- Ask before destructive commands, irreversible migrations, privilege changes, or external publication.

## Completion
- Keep the change bounded to the requested scope.
- Summarize modified files, validation evidence, remaining risks, and any manual follow-up.

Source evidence

AGENTS.md open formatofficial