Repository navigation
Claude won't even read the repo without flagging safeguards, does this actually work? #174
Replies: 1 comment 1 reply
|
Hey @JILTechnology — that's expected, and it's not a bug on either side. Pasting a raw repo URL into a chat gives Claude an offensive-security codebase with zero engagement context, so the safety classifier does exactly what it should and flags it. That's not how the toolkit is meant to run. It's a Claude Code plugin/skill bundle, not something you point a chat window at. Install it, and the skills load into your own authorized workflow: /plugin marketplace add elementalsouls/Claude-BugHunter Then run it inside an authorized engagement (your own target, a bug-bounty program in scope, a CTF). The README's install + usage sections cover the setup, and scope/authorization is enforced by design (/scope pre-flight, read-only defaults) — that's the harness doing its job, not getting in your way. tl;dr: don't feed Claude the link — install the plugin and drive it from a scoped engagement. Works great once it's in the right place. |
Uh oh!
There was an error while loading. Please reload this page.
Has anyone actually used this? I simply copied the link of the repo to it and it flagged the safeguard due to cyber.
All reactions