What happens when my trial ends?

When your 7-day free trial expires, you will need an active subscription to continue using Agent Watch. Your data is preserved, so you can pick up right where you left off once you subscribe.

Yes. You can upgrade from monthly to annual billing at any time. When you switch, we prorate your existing payment so you only pay the difference.

What agents are supported?

Agent Watch supports Claude Code, Codex CLI (including CodexAW), and Gemini CLI out of the box. The hook-based architecture means any tool that emits lifecycle events can integrate with minimal configuration.

Do I need to install anything on my server?

No server-side installation is required. Agent Watch uses a lightweight bridge binary that runs alongside your AI agent on the workstation and reports over HTTPS, so there is nothing to deploy on your infrastructure.

Agent Watch sends real-time notifications the moment an agent needs your input, encounters an error, or finishes a task. You can route alerts to Slack, Microsoft Teams, Discord, email, or SMS — no code required.

Agent Watch vs DIY

Why teams choose Agent Watch over DIY monitoring

Building your own AI agent observability stack sounds simple until you're three sprints deep, wiring webhook receivers, parsing OTLP logs, and duct-taping Grafana dashboards together. Here's how a purpose-built platform compares.

Feature-by-feature comparison

Feature	Agent Watch	DIY / Manual
Setup time	✓ < 10 minutes	✗ Hours to days
Real-time dashboard	✓ Included	✗ Build from scratch
Multi-agent support	✓ Claude, Codex, Gemini	✗ Custom per agent
Alert integrations	✓ Slack, Teams, Discord, SMS	✗ Wire each manually
Token & cost tracking	✓ Automatic via OTLP	✗ Parse logs yourself
Terminal remote control	✓ Built-in SSH panels	✗ SSH + tmux + scripts
Session history	✓ Searchable timeline	✗ Grep through logs
Team management	✓ Multi-user with roles	✗ Roll your own auth
Maintenance	✓ Zero — SaaS updates	✗ Ongoing ops burden

What you'd need to build yourself

A DIY observability layer for AI agents isn't just one project — it's at least three, each with its own maintenance cost.

Custom webhook receivers + database schema

Every agent emits events differently. You'll need bespoke endpoints, a normalized schema, and migration scripts — before you can display a single data point.

Grafana / Datadog dashboards per agent type

Claude Code hooks, Codex CLI callbacks, and Gemini spans each require their own dashboard panels, queries, and alert rules. Triple the config, triple the drift.

Homegrown alerting with retry logic

Want Slack pings when an agent stalls? SMS when a session burns through tokens? You'll be writing queue workers, retry loops, and rate-limit handlers from day one.

Skip the build. Start monitoring in 10 minutes.

Agent Watch gives you a production-ready dashboard for Claude Code, Codex CLI, and Gemini CLI out of the box — no infrastructure required.

Create your free account