Agent skills worth installing.

A disciplined debugging loop that makes an agent establish evidence and isolate root causes before it starts changing code.

Best for

Intermittent failures, unfamiliar codebases, and bugs where the first plausible fix is often wrong.

Keep in mind

The process can feel heavy for obvious one-line defects and still depends on useful logs or reproducible symptoms.

One of the clearest skills for keeping coding agents inside a real red-green-refactor cycle instead of adding tests after implementation.

Best for

Behavior changes and regressions where you need evidence that the test can catch the missing implementation.

Keep in mind

It is intentionally strict and is a poor fit for throwaway exploration, generated artifacts, or projects with no runnable test surface.

A small but high-leverage guardrail that requires fresh command output before an agent claims work is fixed, passing, or complete.

Best for

Long implementation sessions and release work where stale test results or optimistic handoffs create expensive mistakes.

Keep in mind

It improves the evidence behind a claim but cannot determine whether the selected verification command covers the right risks.

A useful counterweight to generic AI interfaces, with concrete direction on visual hierarchy, typography, composition, and intentional design choices.

Best for

Landing pages and product interfaces that need stronger art direction than a default component-library assembly.

Keep in mind

Taste guidance does not replace user research, accessibility testing, or fidelity to an established product design system.

A broad, production-minded React and Next.js review guide drawn from recurring performance problems seen by framework engineers.

Best for

Reviewing React or Next.js changes for waterfalls, bundle growth, rendering costs, and avoidable client-side work.

Keep in mind

Some recommendations are framework-specific and should not override measurements or constraints from the application being reviewed.

Practical database guidance that gives agents a better default for schema design, queries, indexes, row-level security, and operational safety.

Best for

Designing or reviewing Postgres changes in Supabase applications before they become production migrations.

Keep in mind

It cannot inspect production workload characteristics by itself, and Supabase-specific advice may not transfer to every Postgres deployment.