Menu
Use when implementing search with Postgres: tsvector, GIN indexes, search RPC, or replacing client-side filter of large lists. Not for external Algolia/Typesense setup unless migrating away from DB search.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Use when reviewing or fixing accessibility: keyboard navigation, focus order, labels, color contrast, screen reader support, or ARIA attributes. Not for SEO-only or performance-only audits.
Use when adding product analytics (PostHog, GA4, Plausible), event tracking, funnels, or conversion measurement. Not for server logs only or A/B infra unless events are part of the task.
Use when standardizing API and edge function errors: HTTP status codes, error JSON shape, logging, or mapping Supabase/Stripe errors to client responses. Not for UI toast copy only or form field validation messages.
Use when recording audit trails: who changed what, admin action logs, compliance history, or append-only event tables. Not for product analytics (use analytics-events) or debug console logs only.
Use when scheduling background work: cron jobs, digest emails, cleanup tasks, pg_cron, Supabase scheduled edge functions, or job queues. Not for synchronous user-facing API in the request path.
Use when adding dark mode, theme toggle, system theme support, or fixing hardcoded colors that break in dark mode. Not for design-token migration unrelated to dark mode.
| name | postgres-full-text-search |
| description | Use when implementing search with Postgres: tsvector, GIN indexes, search RPC, or replacing client-side filter of large lists. Not for external Algolia/Typesense setup unless migrating away from DB search. |
alter table public.posts add column search_vector tsvector
generated always as (
setweight(to_tsvector('english', coalesce(title, '')), 'A') ||
setweight(to_tsvector('english', coalesce(body, '')), 'B')
) stored;
create index posts_search_idx on public.posts using gin (search_vector);
Use simple config for mixed language products; consider pg_trgm for typo-tolerant name search.
create or replace function public.search_posts(
p_query text,
p_limit int default 20
) returns setof public.posts
language sql stable as $$
select *
from public.posts
where search_vector @@ websearch_to_tsquery('english', p_query)
and deleted_at is null
order by ts_rank(search_vector, websearch_to_tsquery('english', p_query)) desc
limit p_limit;
$$;
Grant execute to authenticated; RLS still applies on underlying table.
const { data } = await supabase.rpc("search_posts", { p_query: q });
Debounce input 300ms; minimum query length 2.
create extension if not exists pg_trgm;
create index projects_name_trgm_idx on public.projects using gin (name gin_trgm_ops);
-- where name ilike '%' || p_query || '%' — use sparingly at scale
ilike '%term%' on large tables without index.soft-delete-and-archiving)