Getting Started with GENE-GoogleSafeBrowsing

Equipping your AI agents with real-time threat intelligence is a straightforward process. By the end of this guide, you will have successfully connected the GENE-GoogleSafeBrowsing tool to your Google Cloud account, stored your credentials securely within your local Feluda environment, and executed your first autonomous URL safety check.

Prerequisites for Setup

Before activating the gene, ensure you have the following ready:

• An active Google Cloud Platform (GCP) account.
• Sufficient permissions to create projects and enable new APIs within your GCP workspace.

1. Provision Your Google Safe Browsing API Key

To communicate with Google's threat databases, your local agent requires a dedicated API key.

1. Log in to the Google Cloud Console.
2. Select an existing project or create a new dedicated project for your security integrations.
3. Using the main navigation menu, go to APIs & Services > Library.
4. Search for Safe Browsing API and click Enable to activate the service for your project.
5. Navigate to APIs & Services > Credentials. Click Create Credentials at the top of the page, then select API key.
6. Copy the newly generated key.

Security Best Practice: Do not leave your key unrestricted. Edit the key's settings in the Google Cloud Console to restrict its usage exclusively to the Safe Browsing API.

2. Securely Store Your Key in Feluda

Because Feluda operates locally on your endpoint, you don't need to pass your API key to external cloud orchestration servers. You will store it directly in your local Feluda secrets manager.

Create a new secret in your environment using these exact parameters:

• Secret Name: apikey_google_safebrowsing (Your AI agent will look specifically for this exact identifier)
• Secret Value: The API key you copied from the Google Cloud Console.

Important: Never hardcode this API key into your prompt templates, workflow scripts, or public repositories. Rely entirely on the Feluda secrets manager to inject the key at runtime.

3. Verify the Connection with a Live Test

With the secret securely stored, it's time to verify that your AI agent can successfully analyze a URL. You will use a standard test URL provided by Google specifically for validating malware detection.

Prompt your AI agent to use the googleSafeBrowsingUrlCheck tool with the following parameters:

• action: check
• url: http://malware.testing.google.test/testing/malware/

Evaluating the Result

If the configuration is correct, the tool will successfully reach Google, identify the test signature, and return a threat_found status. The JSON response cached by your agent should look similar to this:

{
  "input_url": "http://malware.testing.google.test/testing/malware/",
  "normalized_url": "http://malware.testing.google.test/testing/malware/",
  "result": "threat_found",
  "response_format": "json",
  "http_status": 200,
  "key_check": "ok",
  "expressions_count": 5,
  "hash_prefixes_count": 5,
  "full_hashes_count": 1,
  "cache_duration": "300s",
  "cache_key": "sb-1672531200-a1b2c3d4e5f6",
  "cache_path": "/safebrowsing/sb-1672531200-a1b2c3d4e5f6.json",
  "google_response": "{\n  \"fullHashes\": [\n    {\n      \"threatType\": \"MALWARE\",\n      \"hash\": \"...\"\n    }\n  ],\n  \"cacheDuration\": \"300s\"\n}"
}

Troubleshooting Common Connection Errors

If your AI agent fails to return the expected result, check these common configuration hurdles:

• Agent reports missing_secret: The agent cannot find your API key. Double-check that you named the secret exactly apikey_google_safebrowsing in the Feluda secrets manager.
• Agent reports google_api_error (HTTP Status 403): Your API key was rejected by Google. Verify that the Safe Browsing API is actively enabled on your GCP project, and ensure that any API key restrictions you applied are not accidentally blocking the request.
• Agent reports invalid_url: The check action requires a fully qualified URL. Ensure the URL string includes the http:// or https:// protocol prefix.