Real-time logging and observability for AI agents. Track execution, monitor costs, detect anomalies.
AI agents are powerful, but they're also black boxes. Without visibility into what they're doing, you're flying blind:
Existing logging platforms (Datadog, ELK, New Relic) are expensive overkill. Agent Witness is built specifically for agents.
See agent actions as they happen. Search, filter, and explore execution traces in milliseconds.
Monitor token consumption across all agents. Identify expensive operations and optimize before costs spiral.
Detect credential exposure, injection attempts, and suspicious patterns automatically. Fix issues before they become breaches.
Track execution time, latency, and throughput. Find bottlenecks and optimize agent behavior.
Predict token consumption and plan capacity. Know when you'll hit limits before you hit them.
Download logs as JSON or CSV for compliance, audits, or long-term analysis.
Add logging to your agents in under 5 minutes. Works with any framework.
import { AgentWitness } from 'agent-witness';
const logger = new AgentWitness({
token: 'your-api-token'
});
// Log agent actions
logger.log('info', 'Agent started', {
model: 'gpt-4',
task: 'research'
});
logger.log('warning', 'High token usage', {
tokens: 5234,
cost: '$0.15'
});
curl -X POST https://agentwitness.app/api/logs \
-H "Authorization: Bearer YOUR_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"logs": [{
"timestamp": 1708276800000,
"level": "info",
"category": "tool_call",
"message": "Executed web search",
"metadata": {
"tool": "search",
"query": "AI trends 2025",
"duration_ms": 234
}
}]
}'
Perfect for getting started
For teams and scaled deployments
Get instant visibility into your agent deployments. No credit card required for the free tier.