walterfan/lazy-rabbit-skills/lazy-cpp-dev

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Use when working on C++ code, diffs, crashes, reviews, tests, or modernization where RAII, lifetime, UB, STL, headers, threading, sanitizers, or ISO C++ Core Guidelines-style audits matter. Do not use for non-C++, pure C, deploy/infra, or build-only issues without C++ source semantics.

Install

npx vskill@latest install walterfan/lazy-rabbit-skills/lazy-cpp-dev

SKILL.md

lazy-cpp-dev

Practical modern C++ help with a pitfall-first review mindset. Keep diffs small, preserve repo style, prefer RAII and value semantics, treat ownership and lifetime as first-class, and verify with the repo's real build/test path.

References

APIs, headers, security, build/test tooling.

atomics, threads, callbacks, coroutines, lifetime, TSan.

checklist.

guidance.

When To Use

  • User has a C++ repo, diff, file, stack trace, or snippet and wants code-

focused help.

  • Work touches classes, headers, templates, STL, smart pointers, ownership

transfer, public APIs, exception safety, concurrency, tests, or sanitizers.

  • User mentions UB, crashes, leaks, dangling references, slicing, races, data

corruption, std::move, string_view, span, mutexes, coroutines, or "why does C++ do this".

  • Repo has C++ conventions in AGENTS.md, man/, .clang-format,

.clang-tidy, CI, or nearby files that should steer the change.

  • User asks for ISO C++ Core Guidelines-style findings or rule IDs; handle

them directly in this skill.

When Not To Use

  • Non-C++ primary work, pure C, deployment, infra, Jira/GitLab workflow, or

product ops.

  • Build-system-only dependency/linker work unless C++ source semantics are

part of the fix.

  • General C++ history, ecosystem comparison, blog/tutorial writing, or broad

architecture docs without a concrete code artifact.

  • A deliberately non-idiomatic experiment where the user explicitly accepts

the tradeoff.

Pitfall Hotspots

Prioritize these C++-specific bug patterns before style:

  • Naked new/delete, raw owning pointers, mismatched new/delete[], and

leaked resources.

  • Returning pointer/reference/view to a local; dangling string_view, span,

iterator, or pointer after move, mutation, reallocation, or scope exit.

  • Use-after-move: reading a moved-from object before reassignment.
  • Slicing, missing virtual/protected base destructor, and virtual calls from

constructors/destructors.

  • Rule-of-Five / Rule-of-Zero gaps; custom destructor without copy/move

decision; non-noexcept move or swap.

  • Container invalidation after push_back, insert, erase, resize, or

rehash.

  • Uninitialized values, narrowing, signed/unsigned arithmetic, overflow, and

division by zero.

  • C-style casts, unsafe const_cast, type punning, and raw memory operations

on non-trivially-copyable types.

  • Header hygiene: missing guards, global using namespace, unnamed namespace

in a header, cyclic includes, and order-dependent includes.

  • Concurrency: data races, naked locks, volatile as synchronization, locks

across callbacks/co_await, detached threads, and escaping reference captures.

  • Unsafe C interop and sensitive logging: strcpy, sprintf, system() with

input, raw JWTs, secrets, PII, or raw response bodies.

Contract

  • Read build files, lint/format config, nearby code, tests, and docs first.
  • Prefer the repo's standard, error model, threading model, types, logging,

and test framework over generic defaults.

  • Make the smallest useful change.
  • Handle ISO C++ Core Guidelines rule-ID audits here; cite IDs only when

confident, otherwise label findings as guideline-aligned.

  • Never log secrets, raw tokens, raw JWTs, sensitive PII, or private response

bodies in examples, tests, logs, or final output.

Workflow

  1. Detect context.

Read relevant build files, CI, format/lint config, conventions, nearby code, and tests. Identify the standard, flags, compiler, build command, test framework, exception/error model, logging, custom types, and threading choices.

  1. Pick the mode.

Use one primary mode unless the user asks for several: implement, review, test, explain, modernize, or tooling.

  1. Apply C++ defaults.

Match repo style, prefer RAII/value semantics, initialize everything, keep APIs explicit, validate boundaries, reuse local utilities, and keep diffs small.

  1. Run the pitfall pass.

Use the hotspots above, and read the focused references when concurrency, modernization, testing, or project-stack details matter.

  1. Verify honestly.

Run nearest relevant repo commands first, then broader checks if justified. Use ASan/UBSan/TSan when supported and relevant. State exactly what ran, failed, or was not run.

Response Shape

  • For implementation: summarize changes, assumptions, risks, and verification.
  • For review: findings first, ordered by severity: UB/lifetime/race/leak,

security, correctness, API/ABI, missing tests, style.

  • For explanation: cover purpose, key components, ownership/lifetime, control

flow, dependencies, concurrency, and important risks.

  • For Core Guidelines-style audits: cite rule IDs only when confident; otherwise

say "guideline-aligned" and explain the concrete C++ risk.

Hard Gates

  • No new raw ownership, dead views/references, use-after-move, slicing,

missing polymorphic base destructors, throwing destructors, unsafe header namespaces, raw locking where RAII works, volatile synchronization, or detached threads using stack-owned data.

  • Final responses state assumptions, findings/changes, risks, and verification.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/walterfan/lazy-rabbit-skills/lazy-cpp-dev/badge)](https://verified-skill.com/skills/walterfan/lazy-rabbit-skills/lazy-cpp-dev)

Something wrong with this skill or its page? Tell us.