SUPERWISE Sentinel

Your team is pasting customer data
into ChatGPT and Claude.

Sentinel catches secrets, PII and internal data on the way out — on the command line, in the apps you build, and in the browser. Inspected on your machine, redacted before it leaves.

Download for Windows ↓Code signing is being finalized.
Click More info, then Run anyway.

Installation for Mac / LinuxThree commands, right here

Free for 30 days. Read the Windows installation guide.

what Sentinel does to a prompt

What was typed

Here's the failing record:
name Dana Whitfield
email d.whitfield@acme.com
key sk-proj-abc123

What the model received

Here's the failing record:
name <NAME>
email <EMAIL>
key <API_KEY>

The person keeps working. Nothing changes for them except what leaves the machine.

What it catches

The leak isn’t an attack. It’s a Tuesday.

Nobody exfiltrates anything. An engineer pastes a stack trace to debug it faster, and a customer record goes with it.

Secrets and keys

API keys, tokens and connection strings pasted into a prompt while debugging. Replaced before the request leaves.

Personal data

Names, emails, phone numbers and account identifiers inside the logs and records people paste in to get help.

A record of every check

Every inspected interaction lands in the Activity Log — what was checked, what was redacted, and when.

See it live →Run a real prompt through a live gateway. No install, no sign-up.

Where Sentinel works

The tools your team already uses.

Anything running on the machine is covered wherever the machine runs. Chat sites are covered through the browser extension.

What you use	Windows	Mac & Linux	Chrome / Edge
ClaudeClaude Code and claude.ai (setup guide)	Protected	Protected	Protected
OpenAICodex and chatgpt.com (setup guide)	Protected	Protected	Protected
Your own apps and agentsOpenClaw, LangChain, or any framework (setup guide)	Protected	Protected	—
Gemini, Copilot, Grok, Perplexity, DeepSeekIn development	Coming soon

Browser extension is Windows only for now.

How it works

Deploy a sentinel. Route your traffic. See everything.

No security code in every app. A proxy runs on your machine, between your tools and the model, and governs every request that passes through it.

Install Sentinel

One installer on Windows, one command on Mac and Linux. Nothing to wire into your applications.

Your tools find it

Claude Code, Codex, your own agents and the browser extension route through the local gateway automatically.

Inspect, redact, log

Every request is checked on the machine. Secrets and personal data are replaced, and the check is recorded.

The Sentinel console listing six gateways with their guardrail counts, average latency and 24-hour interaction volume, above totals for interactions checked, blocked and remediated. — Every sentinel you run, what it checked, and what it stopped — in one place. Sample data.

Built for two teams

Engineers ship it. Security sleeps.

For engineers

Nothing changes about how you work. No SDK, no wrapper, no code review for a security library. Install it and keep going.

For security

One control point for every AI tool on the machine, and a record of what was inspected rather than an assurance that it was.

Mac & Linux

Three commands and you’re protected.

CLI install for Mac & Linux

# 1 — install the CLI (Python 3.11+)
$ pip install superwise-sentinel-cli

# 2 — deploy a gateway
$ sentinel quickstart

# 3 — prove it redacts, with no model call
$ sentinel test --text "email john.doe@example.com · key sk-proj-abc123"
  ✓ redacted: email <EMAIL> · key <API_KEY>

Requires Docker Desktop

Want the full walkthrough? Read the Mac & Linux installation guide.

You wouldn’t put a network online without a firewall.

Your team is already using AI. Sentinel is the layer that watches what it sends.