Best AI Skills for Development & Code Tools in 2026: 15 Compared
Fifteen Claude Code skills ranked for the day-to-day developer workflow — code review, debugging, git, testing, auth, frontend. obra/superpowers takes the top 3.
Development & Code Tools is the biggest single category in the catalog — 363 skills as of this writing. It's also where the AI coding-tools competition is most savage: every CLI, every IDE, every "agent" startup is racing to ship the same handful of capabilities (code review, debugging, git workflows, test running). In that landscape, the skills that win share one trait: they encode a specific opinion about how the work should be done, not just the mechanic of how to invoke a tool. We sorted all 363 by popularity-plus-quality and picked the top 15.
Quick Pick
receiving-code-review — this is the highest-quality skill in the catalog for the half of code review nobody writes about: how to receive feedback without performative agreement. Pair it with requesting-code-review and you've replaced the two most-skipped soft-skill steps in the dev loop.
What These Skills Actually Do
Development & Code Tools skills sit on the inner loop — the work between "I have a task" and "the PR is merged." They're the skills that fire dozens of times a day on a typical dev's machine. Categorically, this group covers:
- Code review (giving + receiving, both surprisingly different jobs)
- Verification and testing (TDD, test runners, webapp testing via headless browsers)
- Git workflows (worktrees, branch finishing, completion criteria)
- Debugging (systematic, language-agnostic protocols)
- Codebase analysis (architecture audits, commit-pattern profiling)
- Foundational primitives (file search, auth, frontend scaffolding)
The skills that rank high all do one of these things with strong opinions about the right way. The ones that didn't make the cut tend to wrap multiple concerns ("a skill for testing AND deploying AND reviewing") and end up doing none of them well. A useful test: if a skill's name is a verb, it usually beats one whose name is a noun.
How We Ranked
We sorted 15 candidate skills by a composite score:
- Popularity signal — the highest of GitHub stars, install count, or ClawHub download count. Log-scaled so a 100-star skill doesn't get buried under a 100,000-star one if the smaller one is meaningfully better.
- Quality score — when set, a 0–5 rubric that breaks ties within popularity tiers. Roughly 15% of catalog skills carry a quality score today; we surface it in the comparison table when available.
The formula is identical across the entire Best-Of 2026 series, so you can compare apples to apples between categories.
The Top 15
1. receiving-code-review
Skill · obra/superpowers · 167.5K signal · quality 4/5 Use when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable — requires technical rigor and verification, not performative agreement.
The take: Most "code review" skills focus on the reviewer side. This one's antidote to the failure mode AI agents fall into hardest: rubber-stamping reviewer comments. It forces verification of each suggestion before implementing it — surprisingly often, the reviewer is wrong, and a skill that prevents reflexive compliance is more valuable than one that helps you write more thorough reviews.
2. finishing-a-development-branch
Skill · obra/superpowers · 167.5K signal · quality 4/5 Use when implementation is complete, all tests pass, and you need to decide how to integrate the work — guides completion of development work by presenting structured options for merge, PR, or cleanup.
The take: The "90% done" problem made into a skill. Most dev agents are great at writing code and terrible at finishing — closing branches cleanly, deciding squash-vs-merge, cleaning up stale work. This skill is the protocol that prevents the half-finished-branch graveyard that accumulates in every Claude Code project.
3. using-git-worktrees
Skill · obra/superpowers · 167.5K signal · quality 4/5 Use when starting feature work that needs isolation from current workspace or before executing implementation plans — creates isolated git worktrees with smart directory selection and safety verification.
The take: Git worktrees are the secret weapon for AI-assisted parallel work — multiple Claude Code sessions on independent feature branches, never colliding. This skill removes the only friction (the directory naming and safety dance) and unlocks the workflow most teams haven't realized they should be using.
4. Webapp Testing
Skill · ComposioHQ/awesome-claude-skills · 52.6K signal · quality unrated Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.
The take: The breakout skill of the quarter. Composio bundled Playwright into a single skill that lets Claude actually see what it built — screenshots, console logs, DOM state. Before this, the loop for frontend work was "Claude writes code, you check the browser, you describe what went wrong." Now Claude closes the loop itself. The install velocity here suggests this becomes table stakes by Q3.
5. Better Auth
Skill · veeramanikandanr48/better-auth · 43.3K signal · quality unrated Self-hosted auth for TypeScript/Cloudflare Workers with social auth, 2FA, passkeys, organizations, RBAC, and 15+ plugins. Requires Drizzle ORM or Kysely for D1 (no direct adapter). Self-hosted alternative to Clerk/Auth0.
The take: Auth has been the "I'll just use Clerk" decision for two years. Better Auth is the first credible self-hosted alternative that AI agents can wire up correctly in one session — covering the failure modes (CSRF, session fixation, passkey verification) that hand-rolled auth gets wrong. Use when you don't want a Clerk bill, can stomach an extra service to operate, and need RBAC.
6. Agent Skills by Addy Osmani
Plugin · addyosmani/agent-skills · 21.0K signal · quality unrated 20 production-grade engineering skills for AI coding agents covering the full dev lifecycle: define, plan, build, verify, review, and ship — with quality gates and anti-rationalization tables.
The take: A plugin not a skill — meaning you get 20 skills in one install. The "anti-rationalization tables" are the gem: they catch the moments when AI agents try to skip a step ("we don't need tests for this trivial change") with predefined refutations. The closest thing to Google-style engineering discipline you can install in 30 seconds.
7. Code
Skill · ivangdavila/code · 17.9K signal · quality unrated Coding workflow with planning, implementation, verification, and testing for clean software development.
The take: The opinionated "general coding" skill — encodes a plan-then-build-then-verify discipline that prevents the most common Claude Code failure mode (jumping to code before understanding the change surface). Lighter than the Addy Osmani plugin and useful as a single-skill drop-in when you don't want to install 20 at once.
8. Debug Pro
Skill · cmanfre7/debug-pro · 15.6K signal · quality unrated Provides a 7-step debugging protocol plus language-specific commands to systematically identify, verify, and fix software bugs across multiple environments.
The take: Encodes the discipline you wish you'd had on your first debugging job — reproduce, isolate, hypothesize, verify, fix, regression-test, document. The language-specific command library means it knows the right tool for Node vs Python vs Rust without you spelling it out.
9. Code Analysis Skills
Skill · wscats/code-analysis-skills · 14.9K signal · quality unrated This skill should be used when the user needs to analyze Git repositories, compare developer commit patterns, work habits, development efficiency, code style.
The take: Different from the per-PR review skills above — this one profiles an entire repo's history (who commits what, when, in what style). Useful for engineering-manager use cases that the other top skills don't touch.
10. Who Is Actor
Skill · wscats/who-is-actor · 14.0K signal · quality unrated This skill should be used when the user wants to analyze a Git repository and profile each developer's commit habits, work habits, development efficiency.
The take: A focused cut of the Code Analysis Skills above — just the per-developer profile. Useful when you've inherited a codebase and want to understand who knows what without scheduling 1:1s.
11. Eno Skills
Skill · wscats/eno · 13.5K signal · quality unrated Analyze frontend project architecture, evaluate tech stack choices, review component design patterns, and assess build configurations.
The take: Frontend-specific architectural review — strongest fit when you're auditing a React or Vue codebase and need a structured second opinion on stack choices before committing to a refactor.
12. File Search
Skill · xejrax/file-search · 12.6K signal · quality unrated
Fast file-name and content search using fd and rg (ripgrep).
The take: The unsexy primitive that pays off across every other skill. Replaces Claude's default grep calls with rg so symbol-level searches in monorepos return in 50ms instead of 5s. If you work in any codebase larger than 100k lines, install this first.
13. Frontend Design Ultimate
Skill · kesslerio/frontend-design-ultimate · 12.3K signal · quality unrated Create distinctive, production-grade static sites with React, Tailwind CSS, and shadcn/ui — no mockups needed. Generates bold, memorable designs from plain text requirements with anti-AI-slop aesthetic guardrails.
The take: The "no-mockup-needed" claim is what earns the install velocity. The "anti-AI-slop" framing is the discriminator: a deliberate guardrail against the generic-purple-gradient look that plagues AI-generated UI. Worth the install for the design defaults alone.
14. Test Runner
Skill · cmanfre7/test-runner · 11.9K signal · quality unrated Write, run, and manage unit, integration, and E2E tests across TypeScript, Python, and Swift using recommended frameworks.
The take: Multi-language is the differentiator. Most test skills assume one language; this one routes between Vitest/Jest, pytest, and XCTest by inspecting the project. Pairs well with the TDD skill from obra/superpowers — let the TDD skill drive the discipline, let Test Runner pick the framework.
15. Code Review
Skill · wpank/code-review · 11.8K signal · quality unrated Systematic code review patterns covering security, performance, maintainability, correctness, and testing — with severity levels, structured feedback guidance, review process, and anti-patterns to avoid.
The take: The reviewer-side counterpart to #1. Where receiving-code-review prevents reflexive agreement, this one prevents the reviewer's failure mode: nitpicking style while missing security holes. Severity-level structure means reviewers learn to escalate the right issues.
Comparison Table
| # | Skill | Type | Stars / Installs | Quality | License |
|---|---|---|---|---|---|
| 1 | receiving-code-review | Skill | 167.5K | 4/5 | — |
| 2 | finishing-a-development-branch | Skill | 167.5K | 4/5 | — |
| 3 | using-git-worktrees | Skill | 167.5K | 4/5 | — |
| 4 | Webapp Testing | Skill | 52.6K | — | MIT |
| 5 | Better Auth | Skill | 43.3K | — | — |
| 6 | Agent Skills by Addy Osmani | Plugin | 21.0K | — | MIT |
| 7 | Code | Skill | 17.9K | — | — |
| 8 | Debug Pro | Skill | 15.6K | — | — |
| 9 | Code Analysis Skills | Skill | 14.9K | — | — |
| 10 | Who Is Actor | Skill | 14.0K | — | — |
| 11 | Eno Skills | Skill | 13.5K | — | — |
| 12 | File Search | Skill | 12.6K | — | — |
| 13 | Frontend Design Ultimate | Skill | 12.3K | — | — |
| 14 | Test Runner | Skill | 11.9K | — | — |
| 15 | Code Review | Skill | 11.8K | — | — |
FAQ
How is this list different from the category page on aiskill.market?
The category page is a directory: every skill in the category, sortable and filterable. This list is editorial — opinionated, time-stamped (2026-05-17), and ranked. Use the directory when you know what you want; use this when you don't.
Why does the #1 pick have fewer stars than #5?
Stars are one signal among several. The composite score above also includes install counts (which reflect actual usage on aiskill.market) and the optional quality score. A skill with a smaller star count can rank higher if its installs or quality score are strong enough to offset.
Are these all free?
Yes. Every skill in this top 15 is open source — MIT, Apache 2.0, or equivalently permissive. License columns in the comparison table show "—" where the upstream repo hasn't declared one explicitly, but in every case we verified the work is freely usable for both personal and commercial projects.
How do I install one?
Each linked skill page has install instructions. The fastest path is the one-line install via the aiskill.market CLI or by adding the source repo as a Claude Code plugin marketplace.
How often does this list update?
Quarterly. We re-run the generator script against the catalog every three months, diff the rankings, and refresh the commentary. Significant shifts (a new skill entering the top 5, a previous top pick falling out) get flagged in a short addendum at the top.
Should I install all 15?
No. Installing too many skills inflates Claude's context window every session and can dilute the specific ones you actually need. A reasonable starter pack: receiving-code-review (#1), webapp-testing (#4), using-git-worktrees (#3), and one debugging skill (#8). Four skills is plenty for most workflows; add more as you encounter the specific gap each one fills.
Related Categories
- Best AI Skills for DevOps & Deployment in 2026
- Best AI Skills for AI & Machine Learning in 2026
- Skills As Functions: The Naming Discipline — for the design philosophy behind why these specific skills outrank the alternatives
Browse The Full Catalog
Find every skill in this category — including the ones that didn't make the top 15 — at the Development & Code Tools page.
Part of the Best-Of 2026 series. Updated 2026-05-17. Skills sampled from a catalog of ~262 active entries with a combined 743.8K popularity signal across the ranked entries.