Development Claude Skills
8 673 development skills from 759 authors. They write and change code. Half of them fit into 1 830 tokens or less — that is what one costs your context window when the agent loads it. 1 213 ship runnable scripts rather than instructions alone. 42 of them cannot work without an MCP server, most often rube. We also found 1 239 copies of these same skills sitting in other people's repositories — counted once here, not 1 239 times.
8 673 unique 759 authors 10 updated this month 1 368 from vendors
Use this agent when you need to analyze code for design patterns, anti-patterns, naming conventions, and code duplication. This agent excels at identifying architectural patterns, detecting code smells, and ensuring consistency across the codebase. <example>Context: The user wants to analyze their codebase for patterns and potential issues.\\nuser: \"Can you check our codebase for design patterns and anti-patterns?\"\\nassistant: \"I'll use the pattern-recognition-specialist agent to analyze your codebase for patterns, anti-patterns, and code quality issues.\"\\n<commentary>Since the user is asking for pattern analysis and code quality review, use the Task tool to launch the pattern-recognition-specialist agent.</commentary></example><example>Context: After implementing a new feature, the user wants to ensure it follows established patterns.\\nuser: \"I just added a new service layer. Can we check if it follows our existing patterns?\"\\nassistant: \"Let me use the pattern-recognition-specialist agent to analyze the n...
Bootstrap new projects with strong typing, linting, formatting, and testing. Supports Python, TypeScript, and other languages with research fallback.
Use this agent when you need to conduct thorough research on a repository's structure, documentation, and patterns. This includes analyzing architecture files, examining GitHub issues for patterns, reviewing contribution guidelines, checking for templates, and searching codebases for implementation patterns. The agent excels at gathering comprehensive information about a project's conventions and best practices.\\n\\nExamples:\\n- <example>\\n Context: User wants to understand a new repository's structure and conventions before contributing.\\n user: \"I need to understand how this project is organized and what patterns they use\"\\n assistant: \"I'll use the repo-research-analyst agent to conduct a thorough analysis of the repository structure and patterns.\"\\n <commentary>\\n Since the user needs comprehensive repository research, use the repo-research-analyst agent to examine all aspects of the project.\\n </commentary>\\n</example>\\n- <example>\\n Context: User is preparing to create a GitHub issue and wants to foll...
Use when searching text in files, codebases, books, or documents. Use when finding files by pattern, searching large files that are too big to read fully, extracting specific content from many files, or when grep/find is too slow. Triggers on "search for", "find occurrences", "look for pattern", "search in files".
Use when debugging TypeScript/JavaScript bugs by tracing call chains, understanding unfamiliar codebases quickly, making architectural decisions, or reviewing code quality. Extract function signatures and JSDoc without full file reads, trace call hierarchies up/down, detect code smells, and follow data flow. Triggers on debugging, understanding codebase, architectural analysis, signature extraction, call tracing.
Use when designing or reviewing web UI, implementing forms/buttons/inputs, fixing visual hierarchy issues, creating color systems, building layouts, or when interface feels cluttered, hard to read, or users don't know what to click.
Perform exhaustive code reviews using multi-agent analysis, ultra-thinking, and worktrees
>- Reviews repository diffs, PRs, codebases, and completed implementations for material, evidence-backed defects and plan compliance. Use this skill when asked to review, audit, critique, find code issues, or evaluate implemented plan work. Do not use for draft-plan review or implementation unless fixes are explicitly requested.
> Deploys any web app, API, or agent to the internet with a strict pre-deploy audit and a preview → health-check → prod flow. Detects Next, Vite, Astro, Remix, Nuxt, SvelteKit, FastAPI, Flask, Express, MCP server, Claude Agent SDK script, Python worker, Dockerized, and static projects; routes to Vercel, Railway, Docker+SSH VPS, or cloudflared tunnel. Also supports a first-class "run locally" mode. At the start of each cloud deploy the skill asks the user whether to run full-auto (audit → preview → prod in one sequence) or step-by-step (confirm each phase); the user can skip that question by saying "auto" or "step by step" up front. Use when the user says "/all-deploy", "/deploy", "/despliega", "deploy this", "ship this", "push to prod", "get this online", "deploy my agent", "deploy my FastAPI", "run it locally", "corre esto localmente", "deploy auto", "deploy step by step", "paso a paso", "how do I put this on the internet", or otherwise asks to publish a project to a URL or run it locally.
> cheaper way?" Triggers when the request involves generating data or fixtures (lists, datasets, sample records), implementing a problem that is likely already solved by a stdlib function, package, or public API (validation, parsing, lookups, auth, date/currency/geo data), or any implementation expected to exceed ~20 lines. Does NOT trigger when the user has explicitly chosen the approach or library, when the task is under ~10 lines, when fixing a bug in existing code, or for infra/terraform/k8s and DB queries. Run the checklist before writing code, stop at the first question that reveals a cheaper path, and take that path.
> Forces Claude to match output scope exactly to what was requested. Triggers on any bounded coding request — "fix X", "add Y", "write a function/script that Z" — at the moment of writing cases, tests, docstrings, type annotations, CLI flags, config options, one-time abstractions, or refactors of surrounding code. Does NOT trigger when error handling, tests, or validation are the explicit subject of the request, or when the user asked for a complete or production-ready implementation. The best code is code you didn't write.
**AUDIT SKILL** — Comprehensive on-chain Solana program auditor and full-stack security review for ANY programming language. USE FOR: auditing Solana/Anchor programs, reviewing smart contract security, checking for vulnerabilities (missing signers, unchecked accounts, arithmetic overflow, CPI attacks, PDA confusion, type cosplay, reinitialization, flash loan exploits, MEV, governance backdoors, timelock bypass), auditing TypeScript/Python/Go/Java/Ruby/PHP/any language, backend/frontend code review, supply chain safety, operational security (multisig, upgrade authority, deploy process), formal verification and testing quality, logging/monitoring/incident response, data privacy/GDPR/SOC2 compliance, change management, penetration testing methodology, AI/ML security, generating audit reports, running full repository audits. Severity 1-10 scale, 20 micro-checklist domains with 1346 individual verification items, plus 131 known attack vectors, chunked file-by-file execution, item-by-item verdicts. Benchmarked against CertiK (crypto) and EY/SOC2/COBIT (traditional) audit standards. DO NOT USE FOR: writing new features, general coding, non-security reviews.
Use gh CLI to view, triage, fix, link, create, or close GitHub Issues.
| Comprehensive code review for Golang CLI applications. Produces an actionable checklist covering error handling, CLI framework patterns (Cobra/urfave), testing, performance, security, and Go idioms.
Use when modifying a CLI, API, SDK, CI job, workflow, or MCP server that has embedded LLM calls, model/provider config, or a bounded reasoning checkpoint and needs T2A authoring guidance.
Complexity-management design principles — deep modules, information hiding and leakage, module boundaries, dependency direction, strategic vs. tactical tradeoffs. Triggered by planner-agent, plan-reviewer, implementer-agent, implementer-expert-agent, and code-reviewer before any design, implementation, or review work.
TDD cycle and code quality guardrails — failing-test-first, evidence, dependency, error-handling, and test-quality rules. Triggered by /speq-implement, implementer-agent, implementer-expert-agent, and code-reviewer before any implementation or review work.
Semantic code navigation and editing via Serena MCP — symbol-level tools preferred over grep/find/manual edits. Triggered by /speq-mission, /speq-implement, planner-agent, implementer-agent, implementer-expert-agent, code-reviewer, and recorder-agent.
Query specs via the speq CLI — semantic search, feature listing, structure validation. Use before reading any spec file; every speq orchestrator and sub-agent invokes this first for spec discovery, search, or validation.
Read-only git guardrails and the Conventional Commits reference — no history writes, verify changes before completing. Triggered by planner-agent, implementer-agent, implementer-expert-agent, and recorder-agent; git-agent alone is exempt and uses /speq-git-operations instead.
The git/gh operation-to-command mapping and return formats for git-agent — the sole write-permitted actor in this system. Triggered by git-agent.
>- Design and scaffold an autonomous "loop" around a coding agent for a recurring workflow. Use when the user wants to stop hand-prompting an agent and instead build a system that discovers work, delegates it to a maker, verifies the result with a separate checker, remembers progress on disk, and re-runs until a goal is met or it stops for an honest reason. Triggers on "build a loop", "loop engineering", "automate this agent workflow", "design an agentic loop", "/goal", "/loop", "Ralph loop", "maker/checker", "run this overnight", "verifier rubric", or "loop contract". Works with Claude Code, Codex, Cursor, and OpenCode.
Implementing idiomatic, safe, and performant Rust code.
Debugging compiler errors and tracing lifetimes.
>- before starting a new phase, when vision drift is suspected, or every 3-5 sessions. Triggers include user saying retro, run a retro, since-last-retro; asks to analyze patterns, recurring mistakes, or correction trends across multiple sessions; says how did we get here or project history in a project-arc context (not single-file diffs or README edits); or notes that MEMORY.md has drifted from what actually happened across sessions. NOT for single-session review, MEMORY.md cleanup, retro templates, or git-level file diffs.
Python scripts for fetching a single HN post (hnparse.py) and searching HN posts via Algolia (hnsearch.py)
Architectural patterns, Entity ORM, Minimal APIs and dependency injection for projects created with Dext Framework (cesarliws/dext).
Pragmatic clean code standards for Delphi — concise, direct, no over-engineering
Guides and standards for using the Intraweb stateful web framework in Delphi projects.
Patterns for developing REST APIs with Horse framework in Delphi
Pragmatic clean code standards for Delphi — concise, direct, no over-engineering
Architectural patterns, Entity ORM, Minimal APIs and dependency injection for projects created with Dext Framework (cesarliws/dext).
Patterns for developing REST APIs with Horse framework in Delphi
Guides and standards for using the Intraweb stateful web framework in Delphi projects.
当用户明确要求"测试代码"、"运行代码审查"或"进行代码自检"时使用。通过多轮 A 轮批判性代码审查 + B 轮代码质量原则检查，系统化发现、记录、修复程序代码中的问题，并将计划/过程/结果统一沉淀到目标代码根目录的 `.bensz-api/task-{yyyymmdd-hhmm}-{简短描述}/auto-test-code/{yyyy-mm-dd-hh-mm}/output/tests/` 隔离工作区。⚠️ 不适用：用户只是想优化功能（应直接修改）、只是询问代码问题（应直接回答）、没有明确"测试代码"意图。
当用户明确要求"使用 brainstorming"或"使用 awesome-code"时使用。⚠️ 不适用：用户只是想优化/改进某个功能（应直接修改）、只是询问技能问题（应直接回答）、没有明确使用 brainstorming/awesome-code 的一般性开发。
当用户明确要求"使用 awesome-code / 多代理协作 / 并行协调开发"时使用。通过脚本收集可用 Agent 摘要、配置约束与 `dispatch_gate`，再由 AI 自主判断 single-pass / focused-agent / parallel / sequential 策略并选择子代理；当配置中的 required route agent 缺失时必须阻塞继续执行。⚠️ 不适用：用户仅需单一角色的简单修改或咨询、用户未明确表达多代理协作意图、用户只是了解技能概念。
后端开发专家。精通 Node.js/Python/Go/Rust 等后端技术栈，专注于 API 设计、数据库优化、认证授权、微服务架构和性能调优。用于后端服务开发、API 设计和系统架构。
Git 工作流专家。规范化版本控制，确保提交历史清晰可追溯。支持 Conventional Commits 规范、Pull Request 最佳实践、分支管理策略和自动化工作流。
前端开发专家。用于前端应用、页面、组件、现有 UI 增量改进、已有页面视觉优化、UI/UX 改造、视觉重设计、响应式布局、表单和输入框视觉一致性、设计系统落地、React/Vue/Next.js/Nuxt、Tailwind、shadcn/ui、Radix、MUI、AntD、仪表盘、SaaS、Landing Page、移动端适配、前端性能优化、浏览器截图检查和去除 AI 味 UI。触发后应把“做好看”转化为可执行设计约束，优先复用项目现有组件与视觉语言，小步实现并尽量通过浏览器或 Playwright 做视觉反馈闭环；当用户要改进当前页面或已有界面时，默认先诊断再做局部优化，而不是从 0 重做。
当用户明确要求"优化镜像源"、"配置国内镜像"、"加速部署"、"切换镜像源"或项目涉及 Docker/Python/Node.js/Go/Java/Ruby/Rust 等需要包管理器的场景时使用。自动检测项目使用的包管理器，生成适配的国内镜像源配置，确保在国内网络环境下快速部署。⚠️ 不适用：用户只是想了解镜像源概念（应直接说明）、项目已完成部署且无需优化（应直接告知无需操作）、用户明确要求使用官方源（应尊重用户选择）。
当用户明确要求"提交 Git 改动"、"生成 commit 信息"或"创建 git commit"时使用。仅用 Git 分析改动并自动生成 conventional commit 信息（可选 emoji）；必要时建议拆分提交，默认运行本地 Git 钩子（可 --no-verify 跳过），提交后默认自动 push（可 --no-push 跳过）。
当用户明确要求"初始化项目"、"创建项目指令文件"或"生成 AGENTS.md"时使用。完全自动化：自动检测操作系统默认语言，分析项目目录结构（支持 Python/Web/Rust/Go/Java/数据科学/文档项目等），推断项目类型和用途，一键生成规范的项目指令文档。生成结果包括：AGENTS.md（跨平台通用项目指令，Single Source of Truth）、CLAUDE.md（Claude Code 特定适配，通过 @./AGENTS.md 引用）、README.md（项目介绍与使用方法）、CHANGELOG.md（项目变更记录）、.gitignore（Git 忽略规则，安全优先），并在完整初始化时自动补齐 `docs/` 与 `docs/plans/`。
当用户明确要求"发布项目到 GitHub"、"创建 GitHub Release"或"生成 Release Notes"时使用。智能分析 tag 间历史变化，生成专业且吸引人的 Release Notes，自动创建 GitHub Release。支持首次发布、常规版本、预发布版本（alpha/beta/rc），自动识别 prerelease 标记。
Detect Frappe project type, installed apps, version, and tooling. Use as the first step when working on any Frappe/ERPNext codebase to understand the project structure before making changes.
Build REST and RPC APIs in Frappe including whitelisted methods, authentication, and permission handling. Use when creating custom endpoints, integrating with external systems, or exposing business logic via API.
Scaffold and architect custom Frappe apps including app structure, hooks, background jobs, service layers, and production hardening. Use when creating new apps, setting up app architecture, or implementing cross-cutting patterns like caching, logging, and error handling.
The ONE parameterized score-fuser for EVERY comic-author authoring gate — `--gate intent|outline|asset|storyboard|blueprint|continuity|p0_proof|compile`. A single fuser (not a per-layer split) prevents drift. It NEVER re-runs a reviewer; it collects the reviewer score-nodes already on the wiki (via `reviews` edges), fuses them deterministically (min-fuse per dim, max for inverted dims, SKIP missing dims — never substitute 0), then a Codex xhigh adjudicator (NO model pin — follows the local codex config) that sees ONLY structured inputs (scores + tags + raw artifact PATHS + verbatim source context + verbatim rubric — NEVER reviewer prose) makes an asymmetric call (threshold HARD-vetoes "advance"; Codex SOFT-vetoes everything else). The `--gate p0_proof` mode is the zero-credit pre-production proof: a text-only cross-model adversarial review of the pipeline's CODE + IR-CONTRACT + ENGINE state-machine that MUST clear all blockers in BOTH non-author families and then MINT the digest-bound decision:p0_proof certificate via scripts/run_p0_proof.py BEFORE a single metered image-generation credit is spent. Use when a sibling step (intent-parser, outline-creator, asset-review-loop, storyboard-creator, blueprint-author, continuity-audit, json-compiler) defers its acquittal to "the gate", or the user says "过 gate", "cross-layer gate", "审这一层", "p0 proof", "证明流水线再花钱".