Skip to main content

Quick Start

Get AgenticGuard running in under 5 minutes using the proxy URL integration — no SDK required.

Prerequisites​

  • An AgenticGuard account (sign up free)
  • Python 3.9+ (or any language that makes HTTP calls)

Step 1: Create an agent​

Log in to the dashboard, go to Agents, and click New Agent.

Give it a name like my-first-agent and click Create.

Copy the API Key and Proxy URL shown in the success modal.

Step 2: Set your proxy URL​

Your proxy URL looks like:

https://api.agentic-guard.com/proxy/ag_xxxxxxxxxxxx/

Set it as the base URL for your HTTP client:

import os
import httpx

# Replace with your actual proxy URL
PROXY_BASE = "https://api.agentic-guard.com/proxy/ag_xxxxxxxxxxxx/"

# Any URL you prefix with PROXY_BASE is intercepted and checked
response = httpx.get(f"{PROXY_BASE}api.github.com/repos/openai/openai-python")
print(response.json())

AgenticGuard will:

  1. Authenticate your agent by the API key in the URL
  2. Check the call against your policies
  3. Inject any matching credentials automatically
  4. Forward the request and return the response

Step 3: Add a credential (optional)​

Go to Credentials in the dashboard and click Add Credential.

  • Name: github
  • Hostname: api.github.com
  • Type: Bearer token
  • Value: your GitHub personal access token

From now on, any request routed through the proxy to api.github.com will have your token injected — your agent code never touches the credential.

Step 4: Add a policy (optional)​

Go to Policies and click New Policy for your agent.

Paste this YAML to require reasoning for all write operations:

version: "1"
rules:
  - tool: "*"
    action: "POST"
    require_reasoning: true

  - tool: "*"
    action: "DELETE"
    decision: deny
    message: "DELETE calls are blocked by default"

Step 5: Check the audit log​

Go to Audit Log to see every call your agent made, with timing, outcome, and any reasoning provided.

Next steps​