Claude Red

Claude Red

An authorized-use offensive-security methodology reference. It helps plan and organize red-team work for targets you are permitted to test, and points to the public claude-red skill library. It does not ship exploit payloads.

Authorized use only

Use this skill ONLY when you have documented permission: a penetration-testing engagement, an in-scope bug-bounty program, a CTF competition, or an asset you own. The skill's first step is an authorization gate that confirms scope, permission, and rules of engagement before anything else.

What you can do

• Plan an authorized assessment: Scope, recon planning, and methodology selection for in-scope targets.
• Safe validation: Prefer non-destructive proof-of-concept; avoid data loss, denial of service, and irreversible changes.
• Report: Findings with severity, reproduction steps, impact, and remediation.

Hard limits

No out-of-scope or third-party targets, no destructive or mass-automated actions, no persistence, and no use against real systems without authorization. For non-engagement work, use the defensive security-review or anthropic-cybersecurity-skills packages instead.

Upstream

https://github.com/SnailSploit/Claude-Red