Skip to main content

​
Sign up and login

  1. Navigate to console.agentsentinel.dev
  2. Click “Sign Up” or “Sign In” (powered by Clerk)
  3. Create an account using:
    • Email + password
    • Google OAuth
    • GitHub OAuth

​
Quick Start wizard

On first login, you’ll see the Quick Start Wizard - a 4-step guide to get you up and running.

​
Step 1: Welcome

Introduction to Agent Sentinel features:
  • Interventions: See what actions Sentinel blocks
  • Approvals: Human oversight for sensitive operations
  • Runs & Replay: Debug agents with zero-cost replay
  • Activity Ledger: Complete audit trail
  • Analytics: Cost and performance tracking
Click “Next” to continue.

​
Step 2: Create API Key

Generate your first API key for SDK integration:
  1. Click “Generate API Key” button
  2. Copy the key immediately (shown only once)
  3. Store it securely (password manager, secrets vault)
  4. Click “Next”
API keys are shown only once. If you lose it, you’ll need to generate a new one in Settings → API Keys.

​
Step 3: SDK Setup

Choose your language and copy the code snippet: Python: 
from agent_sentinel import enable_remote_sync, guarded_action

# Enable platform sync
enable_remote_sync(
    platform_url="https://platform.agentsentinel.dev",
    api_token="as_your_api_key_here",  # Replace with your key
    run_id="run-001",
)

# Instrument your functions
@guarded_action(name="search_web", cost_usd=0.01, tags=["tool"])
def search_web(query: str) -> dict:
    return {"results": ["..."]}
JavaScript (coming soon): 
import { enableRemoteSync, guardedAction } from 'agent-sentinel';

// Enable platform sync
enableRemoteSync({
  platformUrl: 'https://platform.agentsentinel.dev',
  apiToken: 'as_your_api_key_here',
  agentId: 'my-agent',
  runId: 'run-001',
});

// Instrument your functions
const searchWeb = guardedAction('search_web', { costUsd: 0.01 }, (query) => {
  return { results: ['...'] };
});
Click “Copy Code” and paste into your agent code.

​
Step 4: Complete

Wizard completion with next steps:
  1. ✅ API key created
  2. ✅ SDK setup instructions provided
  3. 🎯 Next: Run your agent and view telemetry in the console
  4. 📚 Learn more: Check out the full documentation
Click “Get Started” to go to the main dashboard.

​
Verify integration

After running your agent with the SDK:
  1. Navigate to Runs page
  2. See your first run appear within seconds
  3. Click the run to see action details
  4. View costs, duration, and outcome
If you don’t see your run within 30 seconds, check:
  • API key is correct
  • platform_url matches your deployment
  • Network connectivity to platform
  • SDK version is up-to-date

​
Explore the console

​
1. Dashboard

Main overview with:
  • 5 stat cards (Total Runs, Units, Consumption, Pending Approvals, Failure Rate)
  • 8 tabbed sections
  • Real-time connection indicator
  • Notifications bell
Try it: Click on stat cards to filter related views.

​
2. Interventions

Policy enforcement monitoring:
  • Total interventions
  • Blocked actions
  • Cost saved
  • High-risk blocks
Try it: Set up a policy with denied actions, then attempt one - see it blocked in real-time.

​
3. Approvals

Human-in-the-loop workflow:
  • Pending approvals with countdown
  • Priority and risk indicators
  • One-click approve/reject
  • Decision history
Try it: Mark an action as requires_human_approval=True and watch it appear in the inbox.

​
4. Runs

Execution history:
  • Run registry with filtering
  • Telemetry view (action sequences)
  • Replay simulation
  • Determinism analysis
Try it: Click “Simulate Replay” on a completed run to test determinism.

​
5. Activity Ledger

Complete audit trail:
  • All actions across all runs
  • Advanced filtering
  • Export capabilities
  • Compliance metadata
Try it: Use the search bar to find specific actions by name or input content.

​
6. Agents

Agent discovery:
  • Auto-registered agents
  • Performance metrics
  • Activity timelines
  • Recent runs
Try it: Click an agent to see detailed statistics and edit display name.

​
7. Analytics

Cost and performance:
  • 30-day consumption chart
  • Daily breakdown
  • Average metrics
  • Time-based filtering
Try it: Switch between 24h, 7d, 30d views to see cost trends.

​
8. Settings

Configuration:
  • Policies: Create budget limits, denied actions, rate limits
  • API Keys: Generate, view, revoke tokens
  • Notifications: Configure Slack, Email, Webhooks (coming soon)
Try it: Create a policy with a $1 run budget, then watch it enforce in real-time.

​
Common tasks

​
Create a budget policy

  1. Go to Settings → Policies tab
  2. Click “Create Policy”
  3. Set name: “Development Budget”
  4. Set run budget: $1.00
  5. Set session budget: $5.00
  6. Enable the policy
  7. Save

​
Generate additional API keys

  1. Go to Settings → API Keys tab
  2. Click “Generate New Key”
  3. Copy the key immediately
  4. Key appears in list with prefix (e.g., as_a1b2c3...)

​
Approve a pending action

  1. Go to Approvals page
  2. Click “Review” on a pending approval
  3. Review action details, inputs, risk level
  4. Click “Approve” or “Reject”
  5. Add optional notes
  6. Confirm

​
Export activity ledger

  1. Go to Activity Ledger page
  2. Apply any desired filters
  3. Click “Export” button
  4. Choose format (JSONL, JSON, CSV)
  5. File downloads immediately

​
View agent statistics

  1. Go to Agents page
  2. Click on an agent
  3. See comprehensive stats:
    • Total runs, actions, cost
    • Success rate
    • Recent activity (24h, 7d, 30d)
    • Approval metrics
    • Intervention rate

​
Keyboard shortcuts

  • / - Focus search bar (where available)
  • Esc - Close modals and dialogs
  • Cmd/Ctrl + K - Command palette (coming soon)

​
Troubleshooting

​
”No runs showing”

  • Verify API key is correct
  • Check SDK is configured with enable_remote_sync()
  • Confirm agent is actually running
  • Check browser console for errors

​
”Real-time updates not working”

  • Check WebSocket connection indicator (top right)
  • Verify no firewall blocking WebSocket (port 443)
  • Try refreshing the page
  • Fallback polling should activate automatically

​
”API key not working”

  • Ensure key starts with as_
  • Copy full key including prefix
  • Check key is active (Settings → API Keys)
  • Verify organization matches

​
”Approval not appearing”

  • Confirm requires_human_approval=True on action
  • Check action hasn’t already completed
  • Verify timeout hasn’t expired
  • Look in “Decision History” tab for expired approvals

​
Next steps

Interventions

Monitor policy enforcement in real-time

Approvals

Set up human-in-the-loop workflows

Activity Ledger

Explore complete audit trails

Settings

Configure policies and integrations

​
See also