Find what's wrong.
Fix it. Prove it.
RedWatch is an AI agent for defenders. It reads your code, logs and configs, finds the vulnerabilities, writes the fix, and runs the tests, asking you before anything changes.
Requires Node.js 18+. Pay as you go: credits never expire.
A security engineer that never gets tired of reading code.
Point RedWatch at a repo, a log dump or a server config. It investigates with real tools, not guesses, and shows its work.
Secure code review
Injection, auth flaws, unsafe deserialization, secrets in source. Every finding gets a severity, a location and a working fix.
Incident response
Feed it auth logs, web logs or an EDR export. It builds the timeline, finds the entry point and drafts containment steps.
Detection engineering
Turn a sample or a technique into Sigma, YARA, Suricata or SIEM queries, then test them against your own data.
Hardening & config review
Dockerfiles, Kubernetes manifests, Terraform, CI pipelines, sshd and nginx configs. It finds the weak defaults and fixes them.
Dependency & secrets audit
Scans lockfiles and history for vulnerable packages and leaked credentials, and tells you exactly what to rotate.
Authorized testing & CTFs
Help with pentests on systems you own or are engaged to test, plus CTFs and training labs, all within the scope you set.
Nothing changes until you say so.
RedWatch reads freely, but every file edit and shell command stops for your approval, with the exact diff or command in front of you.
- See every diff before it's written.
- Approve, deny or auto-approve for the session.
- Ctrl+C interrupts the agent instantly, mid-command.
- Runs locally: your files stay on your machine except what the agent reads for the task.
Running in under a minute.
No containers, no config files, no infrastructure. One key and one command.
Get a key
Create an account, add credits, and copy your API key from the dashboard.
Open dashboard →
Install & log in
Run the installer for your OS, open a new terminal, and paste your key once.
redwatch login
Point it at a problem
Start it in any project folder and describe the job in plain English.
cd my-app && redwatch
Pay for what you use.
Buy credits, spend them per task. No subscription, no seat fees. Credits never expire.
Starter
- $10 of agent credits
- All features included
- 1 API key
Pro
- $33 of agent credits (+10%)
- All features included
- Up to 5 API keys
- Per-key spend limits
Team
- $120 of agent credits (+20%)
- Unlimited API keys
- Per-key spend limits & usage logs
- Priority support
A typical code review of a small service uses well under $1 of credits. See your live balance anytime with redwatch usage.
Questions, answered.
What data leaves my machine?
RedWatch runs on your computer. Only the content the agent reads for your task (file excerpts, command output, your messages) is sent to the RedWatch API to be processed. Nothing is uploaded in bulk, and you can see every file it opens in the session log. See the privacy policy.
Can it break things on my computer?
Not without you. Reading and searching are automatic; writing files and running commands always ask first, showing the exact change. You can deny any step or interrupt with Ctrl+C.
Can I use it for penetration testing?
Yes, on systems you own or are authorized to test, and for CTFs and labs. Attacking systems without authorization is against our acceptable use policy and gets accounts closed.
How does billing work?
You buy credits in the dashboard. Each request the agent makes deducts a small amount based on how much text it processes. When credits run out the agent pauses; top up and it continues. Credits never expire.
What do I need to install it?
Node.js 18 or newer on Windows, macOS or Linux. The installer drops a single file in your home folder and adds the redwatch command. See the docs for details.
Start your first audit today.
Get a key, install in one line, and point RedWatch at the repo you're most worried about.