​

Search API

ProBeya provides a global search endpoint that queries across all entity types — items, projects, and workspaces — within the current organization. Results are grouped by type and limited to 5 per group for fast, responsive command palette integration.

Search uses PostgreSQL case-insensitive substring matching (ILIKE %query%). All three entity type queries run in parallel for minimal latency.

​

Endpoints

​

GET /api/v1/search

Search across items, projects, and workspaces by name.

tRPC: search.query Auth: Bearer token required (scope: read:search) or session cookie Org context: Required

Parameters:

Example:

curl -H "Authorization: Bearer probeya_sk_live_..." \
     "https://acme.probeya.com/api/v1/search?q=calibration"

Response:

{
  "data": {
    "items": [
      {
        "id": "clx9it001",
        "name": "Calibrate pH meter #7",
        "boardId": "clx9bd001"
      },
      {
        "id": "clx9it015",
        "name": "Calibration schedule review",
        "boardId": "clx9bd003"
      }
    ],
    "projects": [
      {
        "id": "clx9pj005",
        "name": "Equipment Calibration Program",
        "slug": "equipment-calibration",
        "workspaceId": "clx9ws001"
      }
    ],
    "workspaces": []
  }
}

Errors:

​

Search Behavior

​

Matching

• Case-insensitive: “OEE” matches “oee”, “OEE”, “Oee”
• Substring: “road” matches “Product Roadmap”, “roadblock”, “crossroads”
• Pattern: %query% (both leading and trailing wildcards)

​

Limits

Each entity type returns a maximum of 5 results. This keeps the command palette responsive and avoids overwhelming users with too many matches.

​

Tenant Isolation

All queries are filtered by the current organizationId — users can never see entities from other organizations. This is enforced server-side regardless of the search query content.

​

Result Fields

The slug and workspaceId/boardId fields enable the client to construct navigation URLs directly from search results without additional API calls.

​

Performance Considerations

The current implementation uses PostgreSQL ILIKE with leading wildcards, which cannot leverage B-tree indexes. For the current scale this is fast enough. At higher volumes, ProBeya can be configured with:

• Trigram indexes: PostgreSQL pg_trgm extension with GIN indexes for substring search
• Full-text search: PostgreSQL tsvector/tsquery for weighted, ranked results
• External search service: Meilisearch or Typesense for sub-millisecond results with typo tolerance

​

Additional REST Endpoints

Two additional read endpoints are available through the REST API but are less commonly used:

​

GET /api/v1/notifications

List notifications for the authenticated user with cursor-based pagination.

tRPC: notifications.list Scope: read:notifications

​

GET /api/v1/dashboard/stats

Get aggregate dashboard statistics for the current organization.

tRPC: dashboard.stats Scope: read:dashboard

No parameters required.