ctf-skills

Agent Skills for solving CTF challenges — web exploitation, binary pwn, crypto, reverse engineering, forensics, OSINT, and more. Works with any tool that supports the Agent Skills spec, including Claude Code.

Installation

npx skills add ljagiello/ctf-skills

Environment Setup

Two setup strategies depending on your workflow:

Pre-install (recommended before competitions)

Use the central installer entrypoint:

bash scripts/install_ctf_tools.sh all

Run a narrower mode when you only want one tool group:

bash scripts/install_ctf_tools.sh python
bash scripts/install_ctf_tools.sh pat
bash scripts/install_ctf_tools.sh apt
bash scripts/install_ctf_tools.sh brew
bash scripts/install_ctf_tools.sh gems
bash scripts/install_ctf_tools.sh go
bash scripts/install_ctf_tools.sh manual

Preview what would be installed (skips already-present packages):

bash scripts/install_ctf_tools.sh --dry-run all

Verify what's already installed:

bash scripts/install_ctf_tools.sh --verify

Use --force to reinstall everything regardless of what's already present. Install logs are saved to ~/.ctf-tools/.

The full package lists now live in scripts/install_ctf_tools.sh.

[!TIP] The pat mode shallow-clones PayloadsAllTheThings into ~/.ctf-tools/PayloadsAllTheThings and links it at ctf-web/payloads/PayloadsAllTheThings. Bulk payloads stay out of git; technique files reference ctf-web/pat-reference.md with a few exemplar payloads inline.

On-demand (during challenges)

Each skill's SKILL.md has a Prerequisites section listing only the tools needed for that category. Install as you go when the agent encounters a missing tool.

Skills

Skill	Files	Description
ctf-ai-ml	3	Model weight perturbation negation, adversarial examples (FGSM, PGD, C&W), foolbox L1BasicIterativeAttack Keras evasion, hand-rolled Keras FGSM via K.gradients, prompt injection, LLM jailbreaking, model extraction, membership inference, neural network collision, LoRA adapter exploitation, gradient descent inversion, data poisoning, backdoor detection, token smuggling, context window manipulation
ctf-web	23	SQLi (blind boolean/time/error generators, second-order, WAF bypasses, EXIF metadata injection, keyword fragmentation bypass, MySQL column truncation, DNS record injection, ORDER BY CASE WHERE bypass, QR code input injection, double-keyword filter bypass), Burp Intruder parity via python-requests.md (sync, 30-worker ThreadPoolExecutor, 100-concurrency httpx async, generator streaming feeds, async_fuzz.py CLI fuzzer), PayloadsAllTheThings on-demand index (pat-reference.md), XSS (AngularJS sandbox escapes, DOM clobbering, CSP bypasses), SSTI (Jinja2, Twig, Go template FuncMap), SSRF (DNS rebinding, cloud metadata), XXE, file upload polyglots, deserialization (Java ysoserial, Python pickle, PHP), JWT (key confusion, dynamic length RSA parameters), OAuth/OIDC, prototype pollution
ctf-pwn	19	Buffer overflow, ROP chains, ret2csu, ret2vdso, vsyscall ROP PIE bypass, bad char XOR bypass, 5 runnable pwntools==4.15.0 template scripts in ctf-pwn/scripts/ with yield payload generators (ret2libc, SROP, shellcraft asm, format string write-size suites, seccomp ORW), heap exploitation (unlink, House of Force, largebin attack guide, glibc 2.39 tcache key protection + calloc behavior, House of Apple 2 + setcontext SUID variant, House of Tangerine/Einherjar), kernel exploitation (io_uring CVE-2024-0582, CET Shadow Stack signal-frame bypass, KASLR, SMEP/SMAP, ret2usr, modprobe_path), format string, SROP with UTF-8 constraints
ctf-crypto	19	RSA (small e, common modulus, Wiener, Boneh-Durfee $d<N^{0.292}$, partial-$d$ exposure, Hastad broadcast, Hastad broadcast with linear padding Coppersmith, Coppersmith structured primes, Franklin-Reiter related message attack e=3, Manger, Manger OAEP timing, Williams $p+1$ + ECM chain, LSB binary search oracle, Fermat/consecutive primes, multi-prime, restricted-digit primes, p=q bypass, cube root CRT, phi multiple factoring, weak keygen base representation, gcd(e,phi)>1 exponent reduction, CRT fault attack, homomorphic decryption oracle bypass, small prime CRT decomposition, Montgomery timing, Bleichenbacher low-exponent forgery, e=1 bypass), AES (ECB byte-at-a-time, CBC bitflip + padding oracle, GCM nonce reuse forbidden attack, ChaCha20-Poly1305 nonce reuse, key-committing partitioning oracle), ECC (small subgroup, invalid curve 5 variants, Smart $p$-adic lift, MOV pairing, twist attacks, SIDH Castryck-Decru warning, CSIDH volcano walk, ECDSA nonce reuse, Ed25519 torsion), classic DH (trivial g, Pohlig-Hellman, Lim-Lee confinement), Lattices/PQC (fpylll Howgrave-Graham lattice, NTRU negacyclic basis, GGH CVP embedding, Mersenne AJPS small roots, BDD predicate, ML-KEM/Kyber flattening), PRNG/stream (PCG XSH-RR, xoroshiro/xoshiro, BBS parity, Henon/Arnold chaos, Spritz/VMPC/RC4A, Trivium cube attack)
ctf-reverse	21	Binary analysis, custom VMs (+ VM bytecode lifting to LLVM IR), WASM, RISC-V, Rust serde, Python bytecode, Unicorn CPU emulation guide (unicorn-emulation.md: hooks, mixed-mode 64→32 retf with XMM state preservation, MMIO peripherals, Keystone trace inversion), Go (GoReSym), Android APK/DEX/JNI, anti-debug, anti-VM, control flow flattening deobfuscation, r2frida live-process instrumentation, MBA simplification with msynth
ctf-forensics	14	Disk/memory forensics (GIMP raw memory dump visual inspection, Kyoto Cabinet hash DB forensics), RAID 5 XOR recovery, APFS snapshot recovery, Windows KAPE triage, Windows/Linux forensics, steganography (Arnold's Cat Map descrambling, MJPEG extra bytes after FFD9, high-res SSTV custom FM demo