AI agent notifications for macOS

Receive AI agent notifications and background results on Mac.

BenAgent gives completed cron, hook, webhook, and plugin work from supported agent connections a visible return path on macOS—without turning BenAgent into the runtime or a hosted prompt relay.

Free download for macOS 14+. Normal voice and agent use requires a one-time US $9.99 Full Unlock; storefront pricing may vary.

Download BenAgent

Why it fits

Delegated work is not useful if you have to keep polling for it.

A prompt response belongs to the active request. A cron or plugin event can arrive later, after the terminal or chat window is gone. BenAgent keeps those paths distinct and gives completed work a local desktop surface.

Receives loopback event posts on the Mac through BenAgent's local listener without an Authorization header.

Supports connection-specific delivery instead of pretending every agent runtime uses the same mechanism.

Keeps event visibility separate from the outgoing Chat Completions request path.

Shows background results beside the workspace where a person can act on them.

Setup path

Wire the event source to the listener you actually intend.

Enable the BenAgent Event Listener

Keep BenAgent running and confirm the listener is ready before configuring a job. The listener is a local inbound surface, not the Hermes or OpenClaw gateway.

Choose the provider-specific delivery path

For OpenClaw, use the BenAgent plugin or supported hook path. For Hermes, configure the cron or webhook delivery target explicitly. Do not copy one provider's setup into the other.

Keep the listener on loopback

The event route has no token authentication. Do not expose the listener beyond the Mac; browser-Origin requests are rejected and payloads still require strict validation.

Test with one harmless job

Trigger a small scheduled task, verify the runtime marks it complete, then confirm the matching result appears on the Mac before relying on recurring automation.

Choose by job

Use runtime logs for diagnosis and BenAgent events for human-visible outcomes.

These surfaces complement each other. Logs explain execution details and failures. BenAgent is for concise results that deserve attention at the desktop edge.

Runtime logs and dashboards

Best for operators who need full traces, retries, plugin diagnostics, schedules, and low-level failure context.

BenAgent desktop events

Best for a person who needs to know that a scheduled or background job finished and see the actionable result without reopening the runtime.

Review the BenAgent OpenClaw plugin

Trust boundary

Loopback events are still untrusted inputs.

Keep the listener local; the event route intentionally has no Authorization header or API key.
Only accepted event shapes that pass input validation should become visible desktop events.
A visible event confirms delivery to BenAgent; it does not replace runtime logs as proof that every upstream step behaved correctly.

Put it into practice