Unified search

One query across tasks, wiki, projects, goals, comments.

Overview

Single search surface powered by a deterministic ranker. Results are returned per-entity type with facet filters, and the ranker upweights exact title matches, recency, and whole-word hits.

Highlights

Cross-entity: task, wiki, project, goal, comment
Kind filter as a query param or facet pill
Deterministic ranker - same query returns the same order every time
Also wired into the ⌘K palette so scope prefixes (`w runbook`) hit /search under the hood

Important to know

Search is case-insensitive and substring-based today; tsvector + BM25 is coming for big workspaces.
Tenant isolation is enforced through the Prisma middleware - no cross-tenant leakage.

How to use it

Type ≥ 2 characters to trigger search.
Click a facet pill to scope the results.
Hit Enter on a hit to open it in the right detail drawer.

FAQ

Why is search slow on very large corpora?

Current implementation is `WHERE title contains ...` - fine up to ~100k rows per entity. Tsvector + GIN index is the next batch.

Atlas docs · Atlas - the AI-native work platform