Every AI Skill in our directory that touches source code — component generators, refactoring tools, test writers, code reviewers, doc writers, migration helpers. Each Skill has been used against real production codebases (not toy examples), tested on multiple languages, and documented for Claude Code, Claude Desktop, Cursor, and Cline. Provider-agnostic — the same Skill runs on Claude, GPT-5, Gemini, Mistral, or DeepSeek.
Development & Code Skills accept a code artifact (a file, a diff, a repo path, a stack trace) and produce a code artifact back — a component, a refactored version, a test file, a review, a doc page, a migration script. The best of these Skills understand your project's conventions rather than starting from generic defaults — they read your existing code first, then generate consistent additions.
The value the Skill layer adds is project awareness. Plain LLM code generation produces code that works in isolation but doesn't match your codebase's patterns — different import order, different naming, different error handling. Skills read the surrounding code first, extract your conventions, then generate code that fits. For teams with a house style, this is the difference between AI code you accept and AI code you rewrite.
What lives here vs. elsewhere: if the Skill's output is source code or an operation on source code, it belongs here. If it's shell/DevOps automation (CI configs, Terraform), see Business Operations. If it's documentation about code that isn't itself code (a technical blog post, a marketing docs page), see Content Creation. For workflows that combine coding with issue tracking and code review across tools, see Custom Workflows.
Our editorial picks — heaviest tested, highest reader ratings.
The gold standard of the category. Reads your project's conventions (naming, imports, styling), then generates components that fit — not generic boilerplate. Supports React 19, Server Components, TypeScript.
Generate unit tests for a file or module. Uses your project's test framework and patterns. Analyzes coverage before and after so you see what's now tested.
AI code review that respects your team's style guide. Flags issues by severity, cites your linter rules, and suggests fixes rather than just complaints.
Filter by sub-type, search by name, sort by recency or alphabetical.
Generate React components that match your codebase — naming, imports, styling, TypeScript. Supports React 19, Server Components.
Sister Skill for Vue 3 Composition API. Reads your project's conventions and generates SFC components.
Given an OpenAPI spec or plain-English requirements, generate Express, FastAPI, or Actix endpoints with validation and error handling.
Generate migration files (Alembic, Prisma, ActiveRecord) from schema change descriptions. Includes rollback logic.
Scaffold new CLI commands (Click, Cobra, oclif) matching your existing command patterns.
Identify chunks of code that should be extracted into functions. Suggests names, parameters, and updates callers.
Incrementally enable TypeScript strict mode. Fixes the resulting errors file by file, or flags cases needing human decision.
Convert React class components to functional components with hooks. Handles state, lifecycle, refs, context correctly.
Identify unreachable code, unused imports, unused variables. Presents as a diff for your review before applying.
Generate unit tests using your project's test framework. Shows coverage before/after so you know what's now tested.
Describe a user flow in plain English; get Playwright or Cypress test code. Includes wait strategies and selector best practices.
For functions with clear invariants, generate property-based tests (Hypothesis, fast-check, PropCheck).
Generate realistic test data fixtures and factory functions matching your model definitions.
AI code review that respects your style guide. Cites specific rules, suggests fixes, groups by severity.
Review a diff or file for security issues: injection, auth bypasses, unsafe deserialization, secrets in code, unchecked user input.
Given a PR diff, generate a summary suitable for the PR description. Explains what changed and why, groups by area.
Generate docstrings for functions, classes, modules. Follows your project's style (Google, NumPy, Sphinx). Includes usage examples.
Scan a repository and generate a README with install, usage, and contribution sections based on what's actually in the code.
Generate reference documentation for a public API from source code + docstrings. Outputs Markdown or MkDocs-ready structure.
Beyond 2to3 — handles harder cases (unicode strings, imports, division) with human-review flags for ambiguous cases.
Incrementally migrate a JS codebase to TS. Adds types file by file, respecting your tsconfig strictness settings.
Generate a GraphQL schema and resolvers from an existing REST API. Preserves auth and validation logic.
Given a stack trace + relevant source, explain the root cause and suggest fixes. Distinguishes symptom from cause.
Scan a log file for anomalies: unusual error rates, new error types, latency spikes, correlated events across services.
MCP servers our team pairs with Development & Code Skills for full workflows.
Read files, browse issues, create PRs. Pairs with every code Skill for end-to-end workflows.
GitLab counterpart for teams on GitLab.com or self-hosted GitLab.
Local file access for reading and writing code in a project directory.
Connect code changes to issue tracking — close issues from PRs, add comments referencing code.
Enterprise issue tracking counterpart for teams on Atlassian.
Read schema for the database migration and API endpoint generator Skills.
Longer-form tutorials for the concepts behind these Skills.
Patterns for making AI code generation feel like it was written by your team.
Empirical guidance from teams running AI review at scale.
Full comparison of the three most-used AI coding surfaces in 2026.
Head-to-head reviews that touch on Development & Code workflows.
Head-to-head on real coding tasks. Verdict: Cursor for tight IDE integration, Claude Code for complex multi-file changes.
Comparison across 5 refactoring scenarios in Python, TS, Rust.
For teams running open-weights locally: which is better in 2026?
Our sister sites cover errors and code recipes across the AI ecosystem. Bookmark them for when a Skill hits an edge case.