# Typesense Index Sync

## Objective

Keep a search index in step with your records so results stay current and correctly scoped.

A one-way sync from the app's records into a Typesense collection, with server-enforced tenant filtering and a rebuild path.

## Before You Begin

This feature is being added to an application that already exists and already
works. Do not scaffold a new project, and do not assume a blank slate.

Inspect the codebase first and establish:

- The existing application structure and where code of this kind already lives.
- The framework and version in use.
- The existing design system — colours, spacing, typography, and component conventions.
- Existing UI components you can reuse instead of writing new ones.
- The existing database structure, if this feature needs to persist anything.
- The existing authentication and authorization system, if this feature is user-scoped.
- Dependencies already installed, so you don't add a library that duplicates one.
- The existing test setup and conventions.

Only start writing code once you understand the above. If the application
already implements part of this feature, extend it rather than replacing it.

## Implementation Instructions

1. Identify every model users search today and declare the indexed shape explicitly: the fields people type into, the fields they filter and sort on, and nothing more. Treat a missing or newly added field as absent rather than letting the document fail validation.
2. Reuse the app's existing background-job system to push creates, updates, and deletes. Do not index inside the request cycle; a slow or unreachable search service must never make a save fail or hang.
3. Build the tenant, workspace, and permission filter on the server at query time and attach it to every search. Do not accept a filter, collection name, or scope value from a client-supplied parameter, because anything the browser can set the browser can change.
4. Support a full rebuild that populates a new collection and only then switches the alias the app reads from, so queries keep hitting the old collection until the new one is complete and verified.
5. Meilisearch Index Sync solves the same problem for a different engine. Choose one engine and route all indexing through it; running both over the same records doubles the write path and guarantees the two drift apart.

## UI and UX Requirements

Match the application's existing design system exactly. Reuse its components,
spacing, and typography. This feature should look like it was always there.

## Responsive Requirements

Works on mobile, tablet, and desktop. Touch targets are large enough to hit on a
phone, and nothing overflows horizontally at 320px.

## Accessibility Requirements

- Fully keyboard navigable.
- Correct semantic elements and ARIA roles.
- Visible focus states.
- Meets WCAG AA contrast.
- Dynamic changes are announced to screen readers.
- Respects prefers-reduced-motion.

## Edge Cases

- The indexed schema will change. Handle optional fields, fields added after the collection was created, and fields that change type, by validating each document against the current shape and quarantining the ones that do not fit rather than aborting the batch.
- Tenant scoping must be applied where the query is assembled on the server. A filter passed through from the client, or merged with a client-supplied filter string, lets one customer read another customer's records.
- Indexing must be idempotent. The same update delivered twice, or a delete that arrives before the create it supersedes, must leave the index in the same correct state, so key documents by the record's stable identifier and carry a version or updated-at value to reject stale writes.
- A rebuild must not blank out search. Index into a separate collection, verify the document count and a sample of queries, then swap the alias, and keep the previous collection long enough to swap back.
- Records too large for the provider's document limit, or carrying values the schema rejects, must be truncated to their searchable text and logged as skipped rather than retried forever in a poison-message loop.
- The search service will be unavailable at some point. Fall back to a plain database query with a visible notice that results may be limited, rather than showing the user an empty result set that reads as no matches.
- Rate limits and bulk import throttling need exponential backoff with jitter and a bounded retry count, so a backlog after an outage does not amplify into a second outage.
- A record that becomes private, is soft-deleted, or moves to a workspace the user cannot see must be removed or re-scoped in the index immediately, since a stale index entry leaks a title and a snippet even when the detail page is protected.

## Testing

Exercise the feature end to end in the running application. Cover every edge case
above, then run the existing test suite and confirm nothing regressed.

## Acceptance Criteria

- [ ] Creating, updating, or deleting a record is reflected in search results without a manual reindex.
- [ ] Indexing runs in the background and a search-service outage never fails or delays a save.
- [ ] Tenant and permission filters are applied server-side and cannot be altered by a client parameter.
- [ ] Replaying the same indexing job produces an identical index state, and stale writes are rejected.
- [ ] A full rebuild completes with no observable search downtime and can be rolled back.
- [ ] Oversized or invalid documents are skipped with a logged reason instead of blocking the queue.
- [ ] When the search service is unreachable, the app degrades to database-backed results with an explicit notice.
- [ ] The feature matches the existing design system.
- [ ] No existing functionality is broken.

## Adaptation Rules

- Match the existing design system. Do not introduce a new colour palette,
  spacing scale, or component library.
- Reuse existing components and utilities wherever they fit.
- Follow the naming, file layout, and code style already present.
- Do not upgrade, replace, or remove existing dependencies to make this
  feature fit. Adapt the feature to the app, not the app to the feature.
- Do not break existing functionality. If a change is genuinely required in
  existing code, make the smallest one that works and say so.
- If something in these instructions conflicts with how the application is
  built, follow the application and explain the deviation.

## Final Verification

Before you report the work as done:

1. Re-read the acceptance criteria above and check each one against what you
   actually built.
2. Run the application and exercise the feature end to end.
3. Run the existing test suite and confirm you have broken nothing.
4. Check the feature on mobile, tablet, and desktop widths.
5. Check keyboard navigation and focus handling.
6. Summarize what changed: files added, files modified, and anything you
   deliberately did differently because of how this application is built.

If any acceptance criterion is unmet, fix it before reporting completion.