Get started

Secure and monitor your AI traffic with Sentinel — in under 2 minutes.

1 - Prerequisites

You'll need Python 3.11+ and pip to install the Sentinel CLI.

Install the CLI via pip

Run the standard install command inside your terminal to download and install the latest release of the Sentinel CLI from PyPI:

pip install superwise-sentinel-cli

Verify the installation by checking the utility version:

sentinel --version

2 - Instrumenting your AI traffic to Sentinel

Choose one of the following deployment paths to connect your runtime traffic environment to a gateway node:

• Option A: Bootstrap a Local Sentinel
• Option B: Connect to a Deployed Sentinel

Option A: Bootstrap a Local Sentinel

sentinel quickstart

The CLI will guide you through the setup process:

1. You will be asked whether you want to use an existing gateway — select N

2. If you haven't authenticated before, you will be prompted for your SUPERWISE® Client ID and Client Secret (required to bootstrap a new gateway). See how to generate credentials.

3. You will be asked which providers you want the gateway to proxy for. To proxy for all of them, just press enter.

4. Once setup completes, proxy settings are automatically configured for your machine.

5. All subsequent sessions will route traffic through this gateway by default.

6. Your newly created Sentinel will appear in the SUPERWISE® SaaS UI

   

Option B: Connect to a Deployed Sentinel

If your team has already initialized an enterprise Sentinel instance (e.g., inside a shared VPC infrastructure network), run the quickstart command with your gateway configuration:

sentinel quickstart

The CLI will guide you through the setup process:

1. You will be asked whether you want to use an existing gateway — select y
2. You will be asked to supply the Sentinel gateway full path.
3. You will be asked which providers you want the gateway to proxy for. To proxy for all of them, just press enter.
4. Once the gateway is configured, proxy settings will be automatically configured for your machine.
5. All subsequent sessions will route traffic through this gateway by default.

3 - Try it out

Run a Security Validation Test

Send a request whose prompt contains a dummy email address and API key. Sentinel redacts both from the prompt before it reaches the model, so the upstream LLM only ever sees {{REDACTED}}. Because the prompt asks the model to echo the data back as a table, you will see the two {{REDACTED}} placeholders reflected in its response — confirming the redaction happened at the edge.

Make requests to different AI provider APIs through your Sentinel proxy. The examples read your upstream provider key from $OPENAI_API_KEY, $ANTHROPIC_API_KEY, or $GEMINI_API_KEY — your own provider keys, not your SUPERWISE® credentials. Export the relevant one first, or paste your key inline:

curl -X POST "$OPENAI_BASE_URL/v1/chat/completions" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -d '{
    "model": "gpt-4o",
    "messages": [
      { "role": "user", "content": "Can you format this data into a clean table?\nID,Timestamp,Actor,Payload\n401,2026-06-30T10:00:12Z,system_job,\"Validating details for user_id:9938\"\n402,2026-06-30T10:01:45Z,sec_audit,\"Triggered email for user [email protected] with key sk-proj-a1B2c3D4e5F6g7H8i9J0k1L2m3N4o5P6q7R8s9T0u1V2w3X4y5Z6\"" }
    ]
  }' | jq -r '.choices[0].message.content'

curl -X POST "$ANTHROPIC_BASE_URL/v1/messages" \
  -H "Content-Type: application/json" \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "claude-opus-4-8",
    "max_tokens": 1024,
    "messages": [
      { "role": "user", "content": "Can you format this data into a clean table?\nID,Timestamp,Actor,Payload\n401,2026-06-30T10:00:12Z,system_job,\"Validating details for user_id:9938\"\n402,2026-06-30T10:01:45Z,sec_audit,\"Triggered email for user [email protected] with key sk-proj-a1B2c3D4e5F6g7H8i9J0k1L2m3N4o5P6q7R8s9T0u1V2w3X4y5Z6\"" }
    ]
  }' | jq -r '.content[0].text'

curl -X POST "$GOOGLE_GEMINI_BASE_URL/v1beta/models/gemini-3.5-flash:generateContent" \
  -H "Content-Type: application/json" \
  -H "x-goog-api-key: $GEMINI_API_KEY" \
  -d $'{
    "contents": [
      { "parts": [ { "text": "Can you format this data into a clean table?\\nID,Timestamp,Actor,Payload\\n401,2026-06-30T10:00:12Z,system_job,\\"Validating details for user_id:9938\\"\\n402,2026-06-30T10:01:45Z,sec_audit,\\"Triggered email for user [email protected] with key sk-proj-a1B2c3D4e5F6g7H8i9J0k1L2m3N4o5P6q7R8s9T0u1V2w3X4y5Z6\\"" } ] }
    ]
  }' | jq -r '.candidates[0].content.parts[0].text'

Now, head directly over to your centralized SUPERWISE® SaaS Dashboard to see your interaction log graphs, validation metrics, and active guardrail latency tables updating live!