The easiest way to ruin a Hack The Box night is to paste the box name into a chatbot and get the whole solution back. You get the root flag and you learn nothing. HackerSkill is our answer to that. It’s a Claude Code plugin that sits next to you while you work a target and coaches you through it instead of spoiling it.

It’ll run the whole lifecycle with you, from recon to foothold to privilege escalation to rooted, and it keeps the notes as you go.

Key Features:

  • Tiered hints. Every hint starts as a nudge. Ask for “more” and you get the technique. Ask it to “spoil it” and you get the answer, but only because you decided you wanted it. You can also say “no spoilers” to lock it to nudges until you say otherwise.
  • It takes the notes for you. Each target gets a targets/<name>/notes.md that acts as the source of truth, with scans in scans/ and loot in loot/. Walk away for a week, run /pwn lame again, and it picks up where you left off.
  • Hybrid execution. The safe, boring enumeration (nmap, gobuster, whatweb, smbclient) just runs. Anything with teeth like exploits, auth attempts, or shells will stop and ask you first.
  • A real playbook. Service by service techniques are built in, and it reaches out to HackTricks, GTFOBins, and exploit-db when a box needs something more specific.

Getting started is two commands in Claude Code:

/plugin marketplace add C33Tech/HackerSkill
/plugin install pwn-challenge@C33Tech-HackerSkill

Then point it at a box and go:

/pwn lame 10.10.10.3

That creates targets/lame/, drops in a notes template, and kicks off the first nmap scan. Run /pwn on its own to list your targets, and /pwn done to wrap up a box once you’ve rooted it.

It shells out to the usual offensive security toolkit (nmap, ffuf, feroxbuster, enum4linux-ng, nxc, evil-winrm, the impacket suite, hashcat, and Metasploit) and it degrades gracefully. Whatever you don’t have installed, it just won’t try to run.

One deliberate constraint: this is for authorized lab, CTF, and engagement targets only. Point it at an IP outside the known lab ranges and it’ll stop and ask you to confirm you have authorization before it does anything. That’s the boring answer, but it’s the right one.

HackerSkill is entirely open source and free to use. If you run it on a box and it coaches you badly, either too much hand holding or not enough, let us know! That feedback is what tunes the hint tiers.