Clean Code Rust
Improve Rust readability and maintainability when the task explicitly involves refactoring, code quality, API clarity, ownership design, error handling, or reducing accidental complexity. Use this skill for Rust cleanup reviews or targeted refactors. Do not use it to fight the crate's established patterns, hide useful explicitness, or broaden a bug fix into an unnecessary rewrite.
Clean Code Js
Improve JavaScript or TypeScript readability and maintainability when the task explicitly involves refactoring, code quality, naming, function design, error handling, or reducing accidental complexity. Use this skill for JS/TS cleanup reviews or targeted refactors. Do not use it to override repo-local conventions, force large rewrites, or apply generic "Clean Code" rules when the task is mainly a…
Doc Write Expert
Expertly write new or review and refresh existing technical and non-technical documentation. Use whenever the user asks to create, draft, document, rewrite, review, audit, verify, or update a README, runbook, guide, proposal, architecture document, ADR, specification, process document, policy, knowledge-base article, AGENTS.md/CLAUDE.md, or other durable prose. Derive claims from authoritative ev…
Code Reviewer
A Claude skill from wnz99/llm-dev-skills.
Cross Review Pr
Cross-model comparative PR / Pull Request review of a PR, branch, commit, or codebase scope. Runs two LLMs through independent reviews, has Reviewer A validate Reviewer B's findings only, then synthesizes overlap, A-only findings, and A-checked B-only findings. Default: Claude <-> Codex. Supports Claude, Codex, and OpenCode via --from/--to. Trigger for comparative PR review, comparative review, P…
Clean Code Py
Improve Python readability and maintainability when the task explicitly involves refactoring, code quality, naming, function design, API clarity, or reducing accidental complexity. Use this skill for Python cleanup reviews or targeted refactors. Do not use it to force heavy OOP patterns, override repo-local conventions, or broaden a bug-fix task into a style rewrite.
Phased Implementation Review Loop
Use this skill whenever the user asks to plan or implement a multi-step code change with subagents, execute work phase by phase, independently check requirement compliance and code quality after every phase, or keep fixing review findings until solved. It unifies implementation-ready planning, fresh implementer subagents, TDD micro-steps, requirement rereads, verification, two-verdict review gate…
Llm Assist
Use this skill instead of ad hoc CLI calls when external LLM help is needed for debugging, code review, planning, root cause analysis, fix verification, rescue after repeated failures, or a second opinion. Default to the complementary model: Codex -> Claude, Claude -> Codex. Ask which provider to use when running inside OpenCode. Use OpenCode only when explicitly requested or approved as a fallba…