agentreflex

Quickstart

Wire your agents, add a reflex, and watch it fire — in a few minutes.

1. Initialize

npx agentreflex init

init detects your agents (Claude Code, Cursor, Gemini CLI, Copilot CLI, Windsurf, OpenCode), asks which to wire, writes each one's native hook, and scaffolds .reflex/ with a starter reflex.

2. Add a reflex

Pull one from the commons:

npx agentreflex add no-secrets

Or write your own. Every reflex is a file in .reflex/, listed in .reflex/config.json:

export default {
  name: "no-force-push",
  onToolCall(ctx) {
    if (ctx.tool === "Bash" && /git\s+push\b.*--force\b/.test(ctx.command ?? ""))
      return { action: "deny", reason: "no force-push — open a PR instead" };
    return { action: "pass" };
  },
};

Edit the file and it takes effect on the next tool call — no rebuild, no restart.

3. Check it

arx dev "git push --force origin main"   # test a command against your reflexes
arx doctor                                # see which agents are enforcing

Next steps

Writing a reflex

The reflex API, in depth.

CLI reference

Every command.

Introduction

agentreflex is the open commons of reflexes for AI coding agents — write an instinct once, and it fires in every agent.

Reflexes

A reflex is a module that reacts to what an agent is about to do and returns a decision.

On this page

1. Initialize2. Add a reflex3. Check itNext steps