CTF AI/ML
CTF reference for ML attacks: model weight tricks, adversarial examples, LLM jailbreaks.
Test report
- Verdict
- Works with setup
- Score
- Tested
- Jul 10, 2026
- Environment
- Claude Code 2.x (agent harness)
- Upstream re-checked
- Aug 10, 2026 · ad752fa
Gave the exact closed-form fix (2xW_orig - W_chal) with runnable code for recovering a suppressed flag, well beyond diffing weights by hand. It's marked non-invocable alone, so installing just this folder may never fire.
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 4/5
What CTF AI/ML does
Quick-reference sub-skill covering AI/ML attack techniques for CTF competitions: model weight analysis (fine-tuning negation, LoRA merging, model inversion), adversarial example generation (FGSM/PGD/C&W), and LLM attacks (prompt injection, jailbreaking, token smuggling). It is one of eight category skills in the ljagiello/ctf-skills repo, meant to be dispatched by that repo's solve-challenge orchestrator rather than triggered fully standalone (frontmatter sets user-invocable: false). Fires on CTF tasks involving ML model files, adversarial robustness, or LLM jailbreak/prompt-injection challenges.
How to install CTF AI/ML
git clone https://github.com/ljagiello/ctf-skills
cd ctf-skills
mkdir -p ~/.claude/skills
cp -r ctf-ai-ml ~/.claude/skills/ctf-ai-ml
Skills live in ~/.claude/skills/ (global) or .claude/skills/
(per-project). Restart Claude Code after installing.
Commands — how to trigger CTF AI/ML
-
/ctf-ai-mlCTF reference for ML attacks: model weight tricks, adversarial examples, LLM jailbreaks.
It also activates on plain-language prompts like these:
-
Craft an adversarial example to bypass this CTF model classifier -
Find a jailbreak prompt injection for this LLM CTF challenge -
Analyze this model's weights for a fine-tuning negation trick
Frequently asked questions
- Is the CTF AI/ML skill free?
- Yes. The skill itself is free from ljagiello/ctf-skills. SkillProof publishes the install command and an independent test verdict at no cost.
- Does CTF AI/ML work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Jul 10, 2026. Verdict: Works with setup. Gave the exact closed-form fix (2xW_orig - W_chal) with runnable code for recovering a suppressed flag, well beyond diffing weights by hand. It's marked non-invocable alone, so installing just this folder may never fire.
- What is the CTF AI/ML SkillProof Score?
- 8.8/10 — installs cleanly 5/5, triggers reliably 5/5, output vs. baseline 8/10, docs & honesty 4/5.
- How do I install CTF AI/ML?
- 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 CTF AI/ML 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.