GETTING STARTED
Use it with any agent
SploitAgent is plain Markdown — there is no runtime to install. You give an agent read
access to the skills/ folder and its operating guide, and it loads the right skill for
each task on its own. Below is the one-minute setup for the common agents.
# Step 1 — get the library
Clone the repository anywhere on your machine.
git clone https://github.com/NoorQureshi/SploitAgent cd SploitAgent
Open your agent inside that folder. It reads the operating guide
(CLAUDE.md for Claude Code, AGENTS.md for everyone else),
which tells it how to pick skills and to confirm scope first.
Give it a target you are authorized to test, and it works the loop — recon, attack, report. That's it.
# Per-agent setup
Claude Code
git clone https://github.com/NoorQureshi/SploitAgent cd SploitAgent claude
Or expose the skills to every project once:
ln -s "$PWD/skills" \ ~/.claude/skills/sploitagent
Codex / Gemini / other CLI agents
git clone https://github.com/NoorQureshi/SploitAgent
cd SploitAgent
codex # or: gemini
They read AGENTS.md automatically. For a one-off, paste a single
SKILL.md into the chat.
Local models (Ollama)
Bake the operating guide into a model so it always knows the method:
printf 'FROM qwen2.5-coder:14b\nSYSTEM """%s"""' \ "$(cat AGENTS.md)" > Modelfile ollama create sploitagent -f Modelfile ollama run sploitagent
Any other agent / framework
Point it at the repo with read access to skills/ and AGENTS.md,
or load the one SKILL.md for the task at hand. There is nothing else to wire up.
# Before you run it
Authorized use only. The first skill every engagement loads is
tradecraft-scope-roe. SploitAgent acts only inside a confirmed envelope — a signed
pentest scope, a bug-bounty program you are in scope for, or systems you own. If scope is unclear
it stops and asks. Read the full rule →