Low-Level Dev Skills - Setup Guide¶
Source: mohitmishra786/low-level-dev-skills (41.5K installs across 142 skills)
GitHub: mohitmishra786/low-level-dev-skills (188 stars)
Skills: 142 skills; top skills: cmake (1.4K), static-analysis (894), llvm (710)
Category: Development / Systems & Low-Level
First Seen: skills.sh Feb 20, 2026; catalogued in the Aug 31, 2026 sweep
Quality Tier: 🟡 Trusted - Gen Agent Trust Hub and Socket Pass, but Snyk carries a warning on the top skill (see Security Audit Status)
A 142-skill suite of procedural guidance for low-level and systems programming, packaged for agents. The catalog is genuinely deep: build systems (CMake, Make, Ninja, Meson), compilers and toolchains (GCC, Clang, LLVM, cross-GCC, MSVC), debuggers (GDB, LLDB), dynamic analysis (Valgrind, sanitizers, heaptrack, fuzzing), performance (linux-perf, flamegraphs, CPU cache optimization, PGO, hardware counters), kernel-adjacent work (Linux kernel modules, eBPF, io-uring, AF_XDP, RDMA, DPDK), assembly (x86, ARM), Rust systems work (unsafe, FFI, sanitizers/Miri, cross-compilation, security), plus Zig, wasm, GPU (CUDA, HIP/ROCm, Triton), embedded (FreeRTOS, OpenOCD/JTAG), and language internals (interpreters, compilers, memory model, branch prediction).
The skills follow a uniform shape: purpose, triggers (example questions that should load the skill), and a numbered workflow - the same trigger-driven pattern Hermes skills use. Hot momentum is current: six skills (rust-cross, linux-perf, clang, compiler-optimizations-deep, zig-debugging, ebpf-rust) were on the skills.sh hot leaderboard during the Aug 31 sweep.
Installation¶
npx skills add mohitmishra786/low-level-dev-skills
Or per-skill:
npx skills add https://github.com/mohitmishra786/low-level-dev-skills --skill cmake
Prerequisites¶
| Requirement | Details |
|---|---|
| Native toolchain | The relevant compiler/build tool for the target skill (gcc, clang, cargo, zig, cmake, etc.) |
| Linux | Most debugging and profiling skills assume Linux (perf, eBPF, Valgrind) |
| Node.js | Recent LTS for the npx skills installer |
Core Skills¶
| Skill | Installs | Use For |
|---|---|---|
| cmake | 1.4K | Modern target-first CMake: out-of-source builds, dependencies, generators, presets, CI/IDE integration |
| static-analysis | 894 | Running and interpreting static analyzers across languages |
| llvm | 710 | LLVM tooling: passes, IR, optimization pipeline |
| gdb | 621 | Interactive and scripted debugging with GDB |
| linux-perf | 517 | Perf profiling, counters, and flamegraph data collection |
| sanitizers | 513 | ASan, TSan, UBSan setup and triage |
| core-dumps | 512 | Core dump analysis and post-mortem debugging |
| clang | 495 | Clang-specific flags, diagnostics, and workflows |
| freertos | 483 | FreeRTOS tasks, scheduling, and embedded patterns |
| linux-kernel-modules | 467 | Building, loading, and debugging kernel modules |
| memory-hierarchy-and-caches | 238 | CPU cache and memory hierarchy optimization (added Sep 29, 2026 sweep) |
Plus assembly-x86, assembly-arm, simd-intrinsics, cpu-cache-opt, memory-model, fuzzing, valgrind, flamegraphs, linkers-lto, binutils, strace-ltrace, dynamic-linking, ebpf, rust-unsafe, rust-ffi, rust-sanitizers-miri, zig-debugging, wasm-emscripten, cuda-debugging, io-uring, riscv-privileged, llvm-passes, mlir, dpdk, triton-lang, and 90+ more.
Security Audit Status (skills.sh, verified Aug 31, 2026)¶
| Skill | Gen Agent Trust Hub | Socket | Snyk |
|---|---|---|---|
| cmake | Pass | Pass | Warn |
Gen Agent Trust Hub and Socket Pass; Snyk carries a warning - the reason this suite ships 🟡 Trusted rather than 🟢 Production.
CorpusIQ Use Cases¶
| Use Case | How |
|---|---|
| Worker-node debugging | The DGX Spark and worker nodes run native binaries, podman, and kernel-adjacent tooling - core dumps, perf, and eBPF skills map directly to that debugging surface |
| Rust and Zig tooling | CorpusIQ's Go/Rust/Python CLI tools get systems-level guidance (unsafe audits, FFI, cross-compilation) when performance work arises |
| Agent-build correctness | CMake, linkers, and sanitizer skills stop agents from guessing build flags when scaffolding native projects |
| Infrastructure forensics | strace-ltrace, flamegraphs, and cpu-cache-opt skills support real diagnosis on degraded nodes |
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.
Limitations / Verification¶
- The top skill sits at 1.4K installs with a long tail below 500; the suite's value is breadth and structure rather than any single blockbuster skill.
- Snyk warning is named in the tier above.
- Skills are guidance, not binaries: the agent must have the underlying toolchain installed.
Verification after install:
npx skills add mohitmishra786/low-level-dev-skills --list # 142 skills discovered
Related¶
- Dart Language Skills Setup
- Skills Catalog - full quality-tiered directory
← Skills Catalog | Skills Marketplace → Powered by CorpusIQ