AutoSugges documentation
AutoSugges turns any list into fast, accurate autocomplete. Publish an immutable, version-addressed artifact once and serve it from the edge, so a keystroke never waits on a database.
Start here
- Create a workspace and let the platform provision your first publishable key.
- Create a list — import CSV, JSON or paste rows, with duplicate and cycle checks.
- Publish it. Publishing compiles an immutable, hash-addressed version to the edge.
- Integrate: pick a framework below, or call the HTTP API directly.
Integration guides
React
Mount the shipped accessible combobox in a React component — debounce, cancellation, keyboard navigation and ARIA are already wired.
Next.js
Keep the input in a Client Component and the page around it a Server Component, so only the interactive part ships JavaScript.
JavaScript (no framework)
Mount an accessible autocomplete on an ordinary HTML page from a data attribute, with no framework and no hand-written combobox markup.
Expo / React Native
Compose the autocomplete from React Native primitives, with the accessibility semantics supplied by the SDK’s native binding.
API reference
Calling the runtime Worker directly — every request parameter, response shape, error code and caching rule, with real curl examples.
Key concepts
- List
- A named, versioned collection with exactly one owner tenant — the thing you publish and query.
- Item
- One entry in a list — a value, a display label, optional metadata and aliases.
- Alias
- A synonym compiled at publish time so a query can match an item without typing its own label.
- Hierarchy
- A parent/child relationship between items — selecting a child can auto-fill its ancestors.
- Version
- An immutable, hash-addressed publish. A client never sees data mixed from two versions.
- Publishable key
- The public, origin-restricted identifier a client uses to query — safe in browser code, never a secret.
- Bucket
- A compiled, pre-ranked set of suggestions for one query prefix — what the edge actually reads.
- Quota
- The query allowance metered at the edge before the cache. Exhaustion returns a structured code, never fake data.