Lesson 6: Tracking Performance (Gas Usage, Uptime, and Reliability in Forta Bots)

Welcome to a critical checkpoint on the monk’s path.

By now, you’ve learned to observe transfers, decode event logs, and emit alerts. But in the real world of production monitoring, detection logic is only half the journey. The other half is resilience.

Today’s lesson focuses on how to build robust, efficient, and reliable bots by monitoring three essential dimensions:

• Gas usage

• Bot uptime

• Execution performance

---

Why Performance Monitoring Is Non-Negotiable

In DevOps and security alike, a broken watcher is worse than no watcher at all. If your Forta bot:

• Runs out of gas

• Fails intermittently

• Emits alerts too slowly

• Or crashes under certain traces

...you lose visibility, and sometimes right when you need it most.

Forta runs your bot against thousands of live transactions per minute across multiple chains. Your agent must be lightweight, stable, and transparent in its own behavior.

---

1. Monitoring Gas Usage

Although Forta bots don’t directly pay gas (Forta nodes do), your bot consumes CPU and memory during its execution.

Each agent has a gas cost equivalent computed by Forta:

• Bots that take longer to execute per tx are more expensive

• If your bot consistently exceeds thresholds, it may be disabled or deprioritized

Tips to Reduce Gas Usage and CPU Cost

• Avoid unnecessary loops

• Don’t parse logs from every contract, better filter by address or topic first

• Don’t decode more data than you need

• Cache reused values (e.g. token decimals)

---

2. Monitoring Bot Uptime

You can track bot uptime on:

Forta Explorer --> Search your bot address --> View performance panel

Key metrics:

• Uptime %

• Number of transactions processed

• Alert frequency

• Average execution time per tx

Forta's backend automatically flags:

• Bots that stop responding

• Bots that error on certain tx types

• Bots with consistently high execution time

---

3. Using handleBlock() for Lightweight Monitoring

If your use case doesn’t require watching every transaction, consider using handleBlock().

Example:

function handleBlock(blockEvent) {
  const findings = [];

  if (blockEvent.blockNumber % 1000 === 0) {
    findings.push(Finding.fromObject({
      name: "Block Milestone",
      description: "1,000 blocks passed",
      alertId: "BLOCK-MILESTONE",
      severity: FindingSeverity.Low,
      type: FindingType.Info
    }));
  }

  return findings;
}

Use this pattern to:

• Emit periodic health signals

• Check state at intervals (like total token supply)

• Reduce computation pressure compared to per-transaction bots

---

4. Using Prometheus + Grafana with Forta

For advanced teams running their own scan nodes, it's possible to integrate Prometheus exporters to monitor:

• Node health

• Agent success/failure rate

• Per-agent metrics: CPU, memory, tx processed

You can also configure alerting thresholds in Grafana dashboards to monitor:

• Latency spikes

• Error rate anomalies

• Sudden drop in alert volume

This goes beyond the basic CLI setup, but it becomes essential when you need to scale.

---

Additional Diagnostic Tips

• Add console logs (visible when testing locally)

• Use try/catch to prevent agent crashes

• Simulate edge cases: no logs, malformed logs, high tx throughput

• Write fuzz tests that push random values into your bot logic

---

Checklist Before Publishing Any Forta Bot

Before publishing a bot make sure to:

• npm run test passes locally

• Handles txs without logs gracefully

• Doesn’t rely on external HTTP requests

• Execution time per tx < 100ms

• Alerts are rate-limited or aggregated if needed

• Metadata fields are short and relevant

• Verified on Forta Explorer under expected chains

---

Example Project to test

detect-large-usdc-transfer-agent

---

Key Concepts Recap

As a Forta Administrator, you are not just building detectors, you are maintaining critical security infrastructure.

Bots that fail silently become blind spots in your organization’s threat detection system.

Mastering gas optimization, uptime monitoring, and performance testing is what turns a good bot into a trusted security asset.

Next lesson, we will explore how to route alerts into systems like Slack, Telegram, or PagerDuty, and set up a true real-time on-call response pipeline.

Until next meditation,
The Blockchain Security Monk