Waxell

Capability · 03 Enforce

For security teams whose people use Claude Code, Claude Desktop and claude.ai

Review each Claude skilland plugin before anyoneinstalls it.

Upload a skill or plugin and Waxell scans it for prompt injection, credential access, data exfiltration and code that runs automatically. Approve what passes, automatically if you choose, and publish it to a private marketplace your Claude clients sync from.

Book a demo

30 minutes. No commitment.

SOC 2 Type II attested · ISO/IEC 27001:2022 certified · HIPAA with a BAA · aligned to the EU AI Act and NIST AI RMF

# what the scanner looks for in every skill or plugin
- prompt injection in instructions
- credential or token access
- data exfiltration (network, clipboard, files)
- code that runs automatically on install

verdict: Safe | Caution | Do not install
  1. 01Discover
  2. 02Observe
  3. 03Enforce
  4. 04Run
  5. 05Prove
What it does
  1. 01

    Scanned before approval.

    Every upload is checked for prompt injection, credential access, data exfiltration and code that runs on install. Findings carry a severity and a verdict.

  2. 02

    Auto-approve the safe ones.

    Set the policy once: approve what passes the scan, route anything with high findings, auto-running code or a do-not-install verdict to a person.

  3. 03

    Published where people already are.

    Approved skills and plugins sync to a private marketplace for Claude Code, Claude Desktop, Cowork and claude.ai. Managed settings lock Claude Code to it. MSPs do this per client through their own GitHub App.

In the recordscan on upload · auto-approval policy · private marketplace · Claude Code managed settings · MSP GitHub App

Also in Enforce

Start with this. The rest of the plane is already there.

Book a demo

30 minutes. No commitment.

SOC 2 Type II attested · ISO/IEC 27001:2022 certified · HIPAA with a BAA · aligned to the EU AI Act and NIST AI RMF