Low-level skills TrothByte

Acknowledgments — Low-level skills TrothByte

This repository is a synthesis. The skills, failure-mode catalogs, and verification disciplines here are built on the work of many open-source projects, standards bodies, and individuals. We are grateful to every one of them.

Skill repositories

The following skill repositories were studied as research input. Where licenses permit, their approaches informed our skill design; content from UNKNOWN-license repositories was used for ideas and methodology only (see LICENSE and registry/claims.yaml for the attribution policy).

Repository	License	How it shaped this work
trailofbits/skills	CC-BY-SA-4.0	Bug-class catalogs (69 Rust / 47–64 C classes), “Rationalizations to Reject”, verification-gate design, ablation-Δ eval methodology, calibration data (PR #238, issues #143/#181/#205/#216)
wshobson/agents	MIT	Survey-style memory-safety and asm pattern skills, PluginEval eval methodology (Monte-Carlo activation, statistical CI)
Jeffallan/claude-skills	MIT	Embedded and C++ RAII/ownership skills
SnailSploit/Claude-Red	MIT	Deep x64 assembly/shellcode content; Windows-on-ARM64 syscall conventions (ideas only — content did not pass the verification bar)
mohitmishra786/low-level-dev-skills	MIT	The broadest coverage map: compilers, linkers, ELF/DWARF, kernel internals (scheduler/MM/VFS), SIMD, ABI, virtualization, HPC
NVIDIA/skills	CC-BY-4.0 / Apache-2.0	GPU programming skills (cuTile/Triton), DOCA/DPU; knowledge-map routing, hardware-safety change discipline, four-source version audit
Z3Prover/z3	MIT	SMT-solver reasoning ground for the smt-z3-sound-usage skill
VoltAgent/awesome-agent-skills	MIT	Index of skill repositories; progressive-disclosure research
zhaoxuya520/reverse-skill	UNKNOWN	Reverse-engineering skill breadth; Go/Rust RE methodology
zhinkgit/embeddedskills	UNKNOWN	Embedded verification discipline (GDB, J-Link as gates)
CSS-Electronics/can-bus-reverse-engineering-skills	MIT	The deterministic CAN protocol RE pipeline, verify-as-gate, SWEEP/HOLDS methodology, parked-vs-moving gate
P4nda0s/reverse-skills	UNVERIFIED (README claims MIT)	Reverse-engineering skill structure; negative-guidance patterns
Masriyan/Claude-Code-CyberSecurity-Skill	MIT	Chaining tables, output-template-first, authorization gates
SimoneAvogadro/android-reverse-engineering-skill	Apache-2.0	Android RE breadth; fingerprint-first triage; R8/Kotlin-metadata name recovery
rudedogg/zig-skills	UNKNOWN	Zig version-pinning problem; WRONG/CORRECT migration pairs (ideas only)
zigcc/skills	UNKNOWN	Zig version policy (VERSIONING.md update-in-place vs new-skill) as idea
nzrsky/zig-skills	MIT	Zig skill structure (derivative of rudedogg)
whit3rabbit/claude-zig-skill	MIT	Zig recipe collection (BBQ Cookbook); multi-version references
0xazanul/fuzz-skill	UNKNOWN	“Proof Standard” for fuzzing claims — the basis of fuzzing-harness-evidence-gate (idea only)
LNC0831/oh-my-fpga	MIT	FPGA/HDL discipline: CDC classification, timing closure, “never make the number green by hiding a violation”
alexfdez1010/risc-v-skill	MIT	RISC-V/RVV intrinsics: VL/AVL strip-mining, LMUL/SEW, tail/mask policies
qwattash/cheri-skills	BSD-3-Clause	CHERI capability semantics (bounds/tags, purecap/hybrid), cheribuild/QEMU harness (author: Alfredo Mazzinghi)
MarsDoge/uefi-firmware-skill	MIT	UEFI/PI spec-boundary discipline, edk2, QEMU serial-log debugging
cellebrite-labs/ghidra-rpc	UNKNOWN	Agent↔Ghidra automation loop (idea only)
manyuegong33/r0crawl_skills	UNKNOWN	RE router / beginner-contract patterns (idea only)
mukul975/Anthropic-Cybersecurity-Skills	Apache-2.0	Breadth reference for security skill taxonomy
hackersifu	MIT	Malware-RE workflows (IOC/unpacking)
rekit	Apache-2.0	RE workflow automation
n132/Libc-GOT-Hijacking	UNKNOWN	Exploit-technique portfolio framing (idea only)

Standards and primary sources

The normative claims in this repository cite the following, among others:

• ISO/IEC WG14 (C11/C23: N1570, N3096) and WG21 (C++20/C++23: N4861, N4971)
• cppreference.com contributors — the C and C++ language references
• SEI CERT (Carnegie Mellon) — the C Coding Standard
• Rust team — The Rust Reference, The Rustonomicon, Rust API Guidelines, Miri
• x86-64 psABI project — System V AMD64 ABI; RISC-V psABI
• Arm — AAPCS64, ABI suite, ARM ARM, CMSIS, NEON/SVE ACLE, Thumb-2
• RISC-V International — ISA specs, V-extension 1.0, RVV C intrinsics
• CHERI project (Cambridge / CTSRD) — CHERI ISA and CheriBSD
• Intel and AMD — SDM and AMD64 APM, Optimization Reference Manuals
• The LLVM project — LLVM LangRef, Clang docs, sanitizer documentation, lld
• GNU — GCC manual, binutils (as/ld/objdump/readelf), GDB manual, GNU Make
• The QEMU project — emulation documentation
• The Linux kernel community — memory-barriers.txt, RCU docs, eBPF docs, scheduler/MM/VFS docs, tracing docs (ftrace/kprobes/kdump), driver API, kbuild, namespaces(7), cgroup v2, OCI runtime spec, syscall tables
• NVIDIA — CUDA C++ Programming Guide, PTX ISA, NCCL; AMD — ROCm/HIP
• InfiniBand Trade Association / rdma-core — verbs documentation; NVIDIA DOCA
• The DWARF committee — DWARF Debugging Information Format v5
• MPI Forum — MPI-4.1; OpenMP ARB — OpenMP 5.x
• UEFI Forum — UEFI & PI specifications; TianoCore — EDK II
• Zig Software Foundation — Zig Language Reference, std source, release notes, build guide
• NASM project — NASM manual; CMake, Ninja, pkg-config documentation
• Devicetree.org — Devicetree Specification; Vector — CAN DBC format; ISO — 11898-1
• MITRE — CWE (incl. CWE-1254) and the CVE list authors
• Frama-C/CEA — ACSL; Diffblue — CBMC; Amazon — Kani; Microsoft Research — Z3, SMT-LIB
• OpenOCD, MCUboot, Espressif ESP-IDF, Zephyr — embedded toolchain documentation
• Reparaz et al. — dudect; Thomas Pornin (BearSSL) — ctgrind; Trail of Bits — constant-time analysis

Talks, articles, and research

The three failure surveys in research/ are grounded in these works, among others:

• Chandler Carruth — Garbage In, Garbage Out: Arguing about Undefined Behavior (CppCon 2016)
• Matt Godbolt — What Has My Compiler Done for Me Lately? (CppCon)
• Sean Parent — C++ Seasoning (GoingNative 2013)
• Aleksey Kladov (matklad) — Preconditions and API design articles
• Herb Sutter — GotW and RAII articles
• Agner Fog — optimization manuals and instruction tables
• Pearce et al. — Asleep at the Keyboard? (IEEE S&P 2022)
• Perry et al. — Do Users Write More Insecure Code with AI Assistants? (CCS 2023)
• Bhatt et al. (Meta) — CyberSecEval (2023)
• Ralf Jung et al. — Stacked Borrows and Rust aliasing semantics
• Mitchell Hashimoto — Ghostty memory leak fix (VM/page-pool analysis, 2026)
• Daniel Lemire — NEON-SIMD per-lane overflow-guard analysis (2026)
• Greg Kroah-Hartman (The Register, 2026) and the LKML review experiments — AI-patch calibration (~⅔ correct)
• Meta — Large Language Model Compiler (arXiv 2407.02524)
• LLM4Decompile (arXiv 2403.05286); FidelityGPT/DeGPT (arXiv 2510.19615); SLaDe (2305.12520); SCDBench (2605.29059); REx86 (2510.20975); capability-cliff (2607.06125); Deconstructing Obfuscation (2505.19887); AutoDecompiler (2606.16162)
• CONCUR (arXiv 2603.03683); RustEvo² (2503.16922); crate hallucinations (2606.08444); We Have a Package for You (2406.10279); MARIN/APIHulBench (2505.05057); RustBrain (2503.02335); crypto-Rust generation (2604.27001)
• ISO-Bench (arXiv 2602.19594); The Correctness Illusion (2606.20128); CommBench (2608.04450); AgentKernelArena (2605.16819); FastKernels (2605.23215)
• NeuComBack (arXiv 2511.01183); SuperCoder (2505.11480); CompilerEval (2511.04132); PeepholeBench (2607.02684); LLVM-Bench (2607.00700)
• LiveFMBench (arXiv 2605.01394); loop-invariant repair (2511.06552); ProVerif/OFMC confidence (2607.20712); The Illusion of Safety (2607.00107); SRE-Bench (2608.11469); Binaries Talk Back (2607.12507)
• EmbedAgent (arXiv 2506.11003); Embedded Arena (2606.16190); FreeRTOS firmware study (2509.09970)

Tools that verified this work

GCC 16.1 (MSYS2), rustc 1.97.1, GDB 17.2, GNU as/ld/objdump/nm, CMake/Ninja, and Python 3.11 were used to execute-verify the source-backed examples in this repository. Without reproducible toolchains, none of the “source-backed” claims would be credible.

---

If your work is represented here and you would like different attribution, or if any attribution is inaccurate, please open an issue and we will correct it promptly.

This site is open source. Improve this page.