Duplicate Detection on Create
Warn users before they create something that probably already exists.
What it adds
A check on create that surfaces likely existing matches and lets the user reuse one instead of adding another copy.
What your agent is told to do
5
What your agent is told to do
5-
1
Pick the fields that actually identify a record — name, email, phone, domain — and match on those rather than on everything.
-
2
Normalize before comparing: trim and collapse whitespace, fold case, strip punctuation, canonicalize phone numbers, and handle email plus-addressing and dots consistently.
-
3
Show possible matches inline before the user commits, with enough detail to tell them apart, and offer to open the existing record instead.
-
4
Recheck on the server at write time. A client-side check loses the race when two people submit at once.
-
5
Do NOT hard-block on a fuzzy match. Similar names are often different people. Let the user override, and record who overrode and why.
Edge cases it handles
6
Edge cases it handles
6- Matching must respect tenant and permission boundaries — never reveal a record the user cannot otherwise see.
- A near-match the user cannot view must be silently ignored, not shown as a redacted teaser.
- Editing an existing record must not flag that record as a duplicate of itself.
- The scan must stay fast on large tables; index the normalized comparison values rather than scanning raw text.
- Genuinely identical names are common in some datasets. If duplicates are legitimate for an object type, do not check it at all.
- An override must be auditable, so someone can later ask why two records exist.
Definition of done
8
Definition of done
8- Likely duplicates are shown before the record is created.
- Comparison values are normalized consistently on read and write.
- The server rechecks at write time and prevents concurrent duplicates.
- Matches are scoped to what the user is permitted to see.
- Users can override the warning and the override is recorded.
- Editing a record never flags it against itself.
- The feature matches the existing design system.
- No existing functionality is broken.
Related features
Weather Widget
Weather Widget
Show current conditions and a short forecast for a place your users care about.
What it does
A compact widget showing current conditions and a few days of forecast for a chosen location.
How it works
- 1 Add this as a tile in the app's existing Customizable Dashboard Widgets rather than as a fixed element bolted into a layout, so users who do not want it can remove it.
- 2 Fetch the forecast on the server on a schedule and serve the widget from the app's own cache. Calling the weather source from every visitor's browser leaks the credential and multiplies the request count by your traffic.
- 3 Show the observation time alongside the reading, and reuse the app's Stale Data Indicator conventions so ageing weather looks the same as any other ageing data in the product.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/weather-widget
Location Map Block
Location Map Block
Put an interactive map on a page showing one location or a whole list of them.
What it does
A reusable map block rendering one or many pinned locations, with a static fallback and a text address alternative.
How it works
- 1 The app already has Mapbox Geocoding and Google Places Address Lookup. Turn addresses into coordinates through the existing integration and store the result on the record; do not add a second geocoding path for this block.
- 2 Make the block take either a single location or a collection, and render the same component for both. A one-pin map is a collection of one, not a separate feature.
- 3 Reserve the block's height before the map initialises so the surrounding page does not jump when tiles arrive. The placeholder should occupy exactly the space the map will fill.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/location-map-block
Record Merge
Record Merge
Combine two duplicate records into one without losing anything attached.
What it does
A merge flow that picks a surviving value per field, reparents all related data, redirects old references, and can be audited afterwards.
How it works
- 1 Show the two records side by side, field by field, and require an explicit survivor for every field where they differ. Do not default to 'newest wins' silently; the user is merging because they know which one is right.
- 2 Reparent every related object — children, comments, attachments, activity, links, memberships — onto the survivor in a single transaction. A half-finished merge is worse than two duplicates.
- 3 Detect relationships that would become duplicates after reparenting (the same person as a member of both records, the same tag applied twice) and collapse them rather than violating a uniqueness constraint mid-merge.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/record-merge
How it works
-
1
Copy the link
Grab the Markdown instruction URL for this feature.
-
2
Give it to your AI
Paste it into Claude Code, Cursor, v0, Lovable — whatever you build with.
-
3
It inspects, then implements
Your agent reads your existing app first, then adds the feature to fit it.
Works with your stack
These instructions are written to adapt. They tell the agent to detect your framework, match your existing design system, and reuse what you already have — rather than assuming a particular stack.
Need it tighter than that? Customize the feature and tell it exactly what you're running.