See what your agent actually does

Know what your AI agents cost, what they touch, and when they go rogue.

Built for OpenClaw. Ready for any agent.

“I built Podwatch after my agent burned $200 in a single weekend without me noticing.”
Rafael Urena, Founder

Join developers who monitor their AI agents.

YOUR LOGS · YOUR MACHINE · YOUR CONTROL

🔒Local-First🛡️Privacy by DefaultOpenClaw Native📡Zero Data Leaves Your Machine

$10/day

What a single agent can cost you daily.

A moderately active agent on Opus burns through ~1M input and ~200k output tokens daily — that's $300/month on API costs alone. Podwatch gives you real-time cost control to prevent budget blowouts.

podwatch --live

2:41psafeRan shell command: git pull origin main
2:41psafeRead file: src/index.ts
2:42pwarnShell command with sudo: npm install -g
2:43pdangerRead file: .env (3 secrets [REDACTED])
2:44p💤 idleHeartbeat check: date, uptime

Your dashboard, at a glance

Real data. Real insights. See exactly what your agents are doing and spending.

Podwatch Costs Dashboard — real-time per-model cost tracking and budget controls

💰 Cost Tracking — per-model, per-session breakdowns

Podwatch Security Dashboard — real-time threat detection and security alerts

🛡️ Security Monitoring — threat detection & alerts

NEWLY SHIPPED

Beyond cost tracking

Full observability for your AI agent fleet — from data flow visualization to session replay.

🔀 Flow Visualization

See how data flows through your agents

Interactive diagram showing channels → agents → models → tools. Spot bottlenecks, track which models each agent uses, and see tool invocation patterns at a glance.

Podwatch Flow Visualization — interactive data flow diagram showing channels, agents, models, and tools
Podwatch Fleet Overview — monitor multiple AI agents with status, costs, and event counts
🤖 Fleet Overview

Multi-agent monitoring

See all your agents in one place — status, costs, event counts, and system info. Know instantly which agents are active and which need attention.

Podwatch Session Viewer — detailed session timeline with costs, models, and event breakdowns
📋 Session Viewer

Session-level detail

Drill into any session to see event timelines, model usage, costs, and duration. Replay what your agent did in every session.

Podwatch Timeline — chronological event feed with filters, risk levels, and tool tracking
📊 Event Timeline

Every action, timestamped

Chronological feed of everything your agent does — shell commands, file reads, API calls — with risk classification and smart filters. Find the needle in 27,000+ events.

Get started in minutes.

No complex setup. No credit card required. Just one command.

1.

Install Plugin

Add Podwatch to your agent's environment.

2.

Connect Agent

Add your API key to start streaming events.

3.

Monitor

Watch your agent's activity in the dashboard.

$ npx -y podwatch

Simple, Fair Pricing

Free forever. Upgrade to Pro for $12/mo when you need more.

$12 / month

Less than 1 hour of unmonitored agent cost.

  • Unlimited Agents
  • Real-time Monitoring
  • Security Alerts
  • Cost & Budget Controls
  • Local-First Audit Logs
  • Community Support
Get Started Free

Free forever. No credit card required.

Frequently Asked Questions

Podwatch uses a local-first architecture. The plugin runs on your machine, collects all data locally, and stores it in an on-disk audit log. Only the data you explicitly configure (like anonymized usage stats or budget alerts) is sent to our servers. Your agent's internal logs and secrets never leave your control.

We are built from the ground up for the OpenClaw framework. Our plugin is designed to integrate seamlessly with its tool and lifecycle hooks. We plan to support other popular frameworks like LangChain and CrewAI in the future.

Yes! The Free plan is forever-free with no credit card required. You get 2 agents, 7-day data retention, unlimited events, cost tracking, and health scores. Upgrade to Pro ($12/mo) anytime for security scanning, budget enforcement, AI analysis, and more.

The data collection plugin is source-available and runs locally on your machine. The dashboard and alert processing systems are managed services. We may explore self-hosted options in the future — contact us if this is important to you.

Privacy-first by design

The plugin collects data locally and transmits only what you allow. Audit logs stay on your machine. The collector plugin is source-available — inspect every line of code.

Stop guessing. Start monitoring.

Gain the confidence to deploy your AI agents into production. Try Podwatch free for 7 days.

Get Started Free