Agent Approve for OpenClaw
Supervise OpenClaw from your phone with guardrails, approvals, and activity monitoring
Agent Approve gives you a dedicated supervision layer for OpenClaw. See what the agent is doing, catch drift early, and require approval before sensitive actions run — all from your iPhone or Apple Watch.
Push notifications keep you informed when OpenClaw is active outside your immediate attention, whether it triggered from a chat message, a cron job, a heartbeat, or an external webhook. You can catch drift, notice suspicious behavior, and stay in control without reopening a terminal.
The same guardrails, activity history, and approval workflow extend across OpenClaw, Cursor, Claude Code, Gemini CLI, Codex, OpenCode, and Copilot.
Quick start
Download Agent Approve from the App Store and complete onboarding.
Run npx agentapprove on your machine and select OpenClaw.
The installer runs openclaw plugins install @agentapprove/openclaw and configures the hooks.
Scan the QR code to pair, then start using OpenClaw.
Install command
npx agentapproveAgent Approve config: ~/.agentapprove/
OpenClaw hook config: ~/.openclaw/openclaw.json
Recommended configuration
- •The installer runs
openclaw plugins install @agentapprove/openclawautomatically. - •For unattended runs, the plugin blocks until Agent Approve returns a decision or the configured timeout is reached.
Capabilities
Good to know
Autonomous supervision – OpenClaw often runs with no human at the terminal. Agent Approve acts as a supervision and guardrails layer: you can see when the agent triggers, what it does, and decide which actions can run automatically versus which ones should require approval.
Cross-channel monitoring – Agent Approve provides a dedicated approval and monitoring workflow for prompts, responses, tool calls, and message-triggered activity — regardless of whether OpenClaw was triggered from Slack, WhatsApp, Telegram, or another channel.
Push notifications – For cron-triggered, heartbeat, or externally invoked runs, Agent Approve sends push notifications so you know when the agent is active.
Drift detection – Especially useful when your concern is not only command approval but agent drift: is OpenClaw staying on goal, being manipulated by incoming messages, or responding in ways you do not want?
Follow-up commands – Supported on the agent_end flow: Agent Approve sends your reply back through OpenClaw's gateway chat API.
Voice input – Especially useful when OpenClaw is running headlessly.
Bulk command parsing – Policy evaluation parses shell commands chained with && into individual sub-commands evaluated independently. A dangerous command like rm -rf hidden as the third step in a chain is caught even when the batch starts with safe commands.
Stay up to date – Run npx agentapprove again to update the plugin, or update directly on your OpenClaw machine with openclaw plugins update @agentapprove/openclaw.
Use cases
Monitor OpenClaw activity from your phone when it triggers from chat, cron, heartbeat, webhook, or another automation surface.
Require approval before OpenClaw executes or sends something sensitive, while still letting safe routines continue automatically.
Catch drift early by keeping tabs on prompts, tool calls, and responses when you are not actively talking to the agent.
Track OpenClaw prompts, tool calls, approvals, and follow-up decisions in one activity history.
Apply the same deny rules to OpenClaw that you use for Cursor, Claude Code, Gemini CLI, Codex, OpenCode, and Copilot.
Questions about OpenClaw
A few common questions about how Agent Approve fits alongside OpenClaw.