OFFLINE SEARCH ENGINE
Sub-millisecond keyword search across years of entries
Find any thought, date, or specific phrase from years ago in milliseconds. Kept runs an embedded SQLite FTS5 search index directly inside your encrypted database.
How it works
Most digital journals rely on remote cloud servers to index and search user text. Kept builds and maintains an external-content Full-Text Search (FTS5) index inside your local SQLCipher database, allowing instant offline search with zero server communication.
Technical implementation
SQLite FTS5 External Content Tables
Search tokens are managed in an optimized FTS5 shadow table synchronized via automatic SQL triggers whenever entries are created, edited, or deleted.
Sub-Millisecond Query Performance
Queries execute in under two milliseconds across thousands of entries, providing instant results as you type in the search bar.
Unicode Tokenizer with Special Character Escaping
Search handles punctuation, special characters, and multi-lingual unicode text robustly without syntax crashes.
100% Offline Indexing
Index updates happen synchronously on device during SQLite transactions, ensuring search remains functional without internet connectivity.
Technical specifications
| Attribute | Specification |
|---|---|
| Search Engine | SQLite FTS5 Extension |
| Indexing Scheme | External content table with SQL triggers |
| Query Latency | <2ms across full archive |
| Network Dependency | Zero (100% local-first) |
Frequently asked questions
Are search queries sent to a remote server?
No. Search runs entirely within the encrypted SQLite database on your device.
Does search include words in attached photos?
Search indexes all text entries, dates, and metadata. OCR for photos is not performed.
Explore other architecture deep dives
Ready to start with Kept?