Sign up
Updated October 2026 · Prices, install paths & star counts re-checked

Best Claude Code Skills in 2026

Six workflow skills that change what Claude Code does — not a 2,000-file bundle. Every pick is previewable before you pay: what you get, who it's for, who it's not for, and the exact command to try it.

See the 6 picksRead the full guide

How we picked these

No fake benchmarks, no invented test scores. Each skill earned its place through at least two of three signals: sustained community adoption (stars and forks, re-checked October 2026), a workflow with a clear before-and-after you can feel in a single afternoon, and depth of content — a real process with steps and checks, not a three-bullet prompt dressed up as a skill.

And unlike most 'best of' lists, every pick here is buyable with a preview: the detail page shows exactly what you get, who it's for, who it's not for, and the command to try it — before you pay.

Debugging

Systematic Debugging

$9

Stop guessing at fixes. A 4-phase root-cause discipline (reproduce → isolate → hypothesize → verify) that forces your agent to find the real cause before touching code.

Try it
/systematic-debugging My checkout API returns 500, but only when two users pay at the same time
Not for:
Greenfield feature work (use test-driven-development); performance profiling (use diagnosing-bugs)
Install:
~/.claude/skills/systematic-debugging/
Source:
obra/superpowers · MIT · ★ 302,815 · Checked Oct 2026
Agents:
Claude Code · Codex · Cursor · Copilot · Gemini CLI
View skill →

Diagnosing Bugs

$9

A battle-tested diagnosis loop for hard bugs and performance regressions: reproduce → minimise → hypothesise → instrument → fix → regression-test.

Try it
/diagnosing-bugs The dashboard gets slower every hour until someone restarts it
Not for:
Obvious bugs (systematic-debugging is faster); writing new code from scratch
Install:
~/.claude/skills/diagnosing-bugs/
Source:
mattpocock/skills · MIT · ★ 299,170 · Checked Oct 2026
Agents:
Claude Code · Codex · Cursor
View skill →

Testing

Test-Driven Development

$9

RED-GREEN-REFACTOR with iron laws: write the test first, watch it fail, then write the minimal code to pass.

Try it
/test-driven-development Add cursor-based pagination to the orders API
Not for:
Throwaway spikes and prototypes; debugging existing failures (use systematic-debugging)
Install:
~/.claude/skills/test-driven-development/
Source:
obra/superpowers · MIT · ★ 302,815 · Checked Oct 2026
Agents:
Claude Code · Codex · Cursor · Copilot · Gemini CLI
View skill →

Review & quality

Code Review & Quality

$9

Multi-axis code review with severity gates (Critical/High/Medium/Low) across correctness, readability, architecture, security, and performance.

Try it
/code-review-and-quality Review the diff on branch feat/refund-flow
Not for:
Line-level AI-slop cleanup (use tidy); full security audits beyond code review
Install:
~/.claude/skills/code-review-and-quality/
Source:
addyosmani/agent-skills · MIT · ★ 109,122 · Checked Oct 2026
Agents:
Claude Code · Codex · Cursor · Copilot
View skill →

Tidy

$9

A diff/PR reviewer built specifically for agent-written code. Catches AI-slop patterns that line-by-line reviews miss, tiers findings as confirmed vs plausible, and can apply the fixes itself.

Try it
/tidy Review my changes and fix whatever you can confirm
Not for:
Creating PRs or triaging CI failures; UX or accessibility review
Install:
~/.claude/skills/tidy/
Source:
mblode/agent-skills · MIT · ★ 151 · Checked Oct 2026
Agents:
Claude Code · Codex · Cursor
View skill →

Git workflow

Git Workflow & Versioning

$9

Branch strategy, commit discipline, PR hygiene, and release versioning for agent-driven development. Keeps AI-generated changes manageable, reviewable, and reversible.

Try it
/git-workflow-and-versioning Split my working tree into clean atomic commits
Not for:
Merge conflict resolution specifics; choosing a git hosting provider
Install:
~/.claude/skills/git-workflow-and-versioning/
Source:
addyosmani/agent-skills · MIT · ★ 109,122 · Checked Oct 2026
Agents:
Claude Code · Codex · Cursor · Copilot · Gemini CLI
Rated 5.0 (1 review) on SkillStack
View skill →

Buy the stack, not just one skill

The six picks are also sold as packs — or unlock everything with a subscription.

Debug Like a Pro

$15

Two complementary debugging disciplines: systematic root-cause analysis plus a battle-tested diagnosis loop with regression testing.

View skill →

Ship Quality

$29

The full quality pipeline for agent-written code: TDD enforcement, multi-axis code review, AI-slop diff review, and disciplined git workflow.

View skill →

Premium subscription

from $9.99/mo

Unlocks every premium skill, prompt and pack — plus all future drops.

See subscription →

Us vs. a $10–$15 bundle

Same question, three checks. Run them on any seller — including us.

See the finished output before paying

Every SkillStack detail page shows what you get, who it's for, and who it's not for. Bundles show a file count.

Named agents, tested

Each skill lists the agents it was tested on and the exact install path for each. Bundles promise 'works with everything' and prove nothing.

If it doesn't work for you

Substantial defects are covered within 14 days of purchase (see Terms). Bundle guarantees usually cover corrupted downloads — not 'it just doesn't work in my setup'.

Why we don't sell 2,000-skill bundles →

Questions, answered

Are Claude Code skills safe?

Skills are markdown instruction files, not executables — they run inside your agent, with your permissions. Still, read before you install: community collections are fully public on GitHub, and every SkillStack premium skill ships its full text in the ZIP after purchase (a teaser is shown before you pay).

Do these skills work in Cursor, Codex, or Copilot?

Yes. The SKILL.md format started with Claude Code but is now read by Cursor, Windsurf, Copilot, Codex, Gemini CLI, and Cline. Each detail page lists the agents that skill was tested on and the install path for each.

Free vs. paid skills — which should I get?

If you want a large corpus to mine for ideas, free collections like Superpowers win — and we'd tell you to get them. Paid skills win when you want one workflow that works out of the box: a single install-ready package, a what-you-get spec, and lifetime updates. You're paying for curation and certainty, not file count.

How do I install a skill I bought?

Download the ZIP and unzip it into your agent's skills folder — for Claude Code that's ~/.claude/skills/<skill-name>/ — then restart your agent. Cursor, Codex, Copilot, and the rest each have their own folder, listed on every detail page. Or connect once via MCP and install without leaving your editor.

What if a skill doesn't work for me?

Digital goods are generally non-refundable once unlocked, but if a skill has a substantial defect, contact us within 14 days and we'll make it right — see the Terms for details. Before buying, check the 'Not for' line on each card: most mismatches are visible there.

Open-source attribution

The six picks are curated, install-ready packages of MIT-licensed open-source work. Systematic Debugging and Test-Driven Development adapt obra/superpowers; Diagnosing Bugs adapts mattpocock/skills; Code Review & Quality and Git Workflow & Versioning adapt addyosmani/agent-skills; Tidy adapts mblode/agent-skills. Every ZIP ships the original LICENSE and an ATTRIBUTION.md. Nothing here hides its upstream.

Start with the one that saves you hours this week

Single skills are $9 each — or take 20% off your first order before the code expires.

First-order code:ZD1UXZDI5FU8

20% off · ends Nov 9, 2026

Browse all skillsSee subscription