Skip to main content
The Audit Log UI and replay feature are planned. The database schema (audit_log, run_events) exists but the dashboard UI is not yet built.

​
Overview

Every agent run is automatically recorded. You can view full audit logs and replay sessions for debugging, compliance, or training.

​
Viewing Audit Logs

// Get run with full audit log
const run = await rt.runs.get('run-abc123', {
  include: ['auditLog'],
});

console.log(run.auditLog);
// [
//   { time: '2025-01-21T10:00:00Z', type: 'start', message: 'Run started' },
//   { time: '2025-01-21T10:00:01Z', type: 'thinking', content: '...' },
//   { time: '2025-01-21T10:00:02Z', type: 'tool_call', tool: 'bash', input: {...} },
//   ...
// ]

​
Audit Log Events

Event TypeDescription
startRun started
thinkingAgent reasoning
tool_callTool invoked
tool_resultTool returned
read_fileFile was read
edit_fileFile was edited
file_editFile was edited
user_inputUser sent input mid-run
pauseRun was paused
resumeRun was resumed
errorError occurred
completeRun completed
cancelRun was cancelled

​
Session Replay

Replay a session step-by-step: 
// Start replay
const replay = await rt.runs.replay('run-abc123');

// Step through events
for await (const event of replay) {
  console.log(event.timestamp, event.type);
  
  // Pause replay
  await replay.pause();
  
  // Resume at any time
  await replay.resume();
  
  // Jump to specific time
  await replay.seekTo('2025-01-21T10:00:30Z');
}

​
Replay Controls

const replay = await rt.runs.replay('run-abc123');

// Speed control
replay.setSpeed(2);    // 2x speed
replay.setSpeed(0.5);  // Half speed

// Jump to event
replay.seekToEvent(15);  // Go to 15th event

// Get current state at any point
const state = await replay.getStateAt('2025-01-21T10:00:30Z');
console.log(state.files);      // Files at that moment
console.log(state.context);    // Agent context at that moment

​
Dashboard Replay

In the dashboard, you can:
  • View timeline of all events
  • Click on any event to see details
  • See file diffs at each point
  • Watch agent “thinking” in real-time replay
  • Export replay as video or transcript

​
Filtering Logs

// Filter by event type
const toolCalls = run.auditLog.filter(e => e.type === 'tool_call');

// Filter by time range
const filtered = await rt.runs.get('run-abc123', {
  include: ['auditLog'],
  auditLog: {
    from: '2025-01-21T10:00:00Z',
    to: '2025-01-21T10:05:00Z',
    types: ['tool_call', 'tool_result'],
  },
});

​
Export

// Export as JSON
const log = await rt.runs.exportAuditLog('run-abc123', {
  format: 'json',
});

// Export as transcript
const transcript = await rt.runs.exportAuditLog('run-abc123', {
  format: 'transcript',
});

// Export as video (coming soon)
const video = await rt.runs.exportAuditLog('run-abc123', {
  format: 'video',
});

​
Compliance Features

For enterprise compliance needs: 
// Set retention policy
await rt.settings.update({
  auditLogRetention: '90d',  // Keep for 90 days
});

// Enable immutable logs
await rt.settings.update({
  immutableAuditLogs: true,  // Cannot be deleted
});

// Enable encryption
await rt.settings.update({
  auditLogEncryption: true,
});

​
CLI Usage

# View audit log
runtools runs log run-abc123

# Replay in terminal
runtools runs replay run-abc123

# Export
runtools runs export run-abc123 --format json > log.json

​
Use Cases

Debugging

Understand what went wrong in a failed run

Compliance

Maintain audit trail for regulated industries

Training

Learn from successful agent interactions

Quality Assurance

Review agent behavior before production