The All-in-One API Kit.
ApexKit collapses the modern backend stack into a single, high-performance binary. Database, Vector Search, AI Runtime, and Edge Logic—unified in Rust.
Hyper-Optimized Rust Runtime
Built on Axum and Tokio. ApexKit isn't just a wrapper; it's a compiled binary that handles high-concurrency workloads with minimal memory footprint. No garbage collection pauses, just raw speed.
Hybrid Relational-Document Store
Powered by LibSQL in WAL mode. ApexKit combines the structure of SQL with the flexibility of JSON document storage, linked by a graph-based _relations table.
Native Multi-Tenancy
Physical isolation per customer. ApexKit manages separate SQLite files and storage folders for every tenant, routed automatically via subdomains.
Embedded Vector Engine
ApexKit integrates Candle (Rust ML framework) to run BERT/Jina embeddings directly on the CPU, coupled with an in-memory HNSW Index for millisecond-fast semantic search. No Pinecone required.
Explore Vector APIs// Generate embedding locally let vec = candle.embed(query); // HNSW Approximate Nearest Neighbor let hits = index.search(vec, 10); // Hydrate results from DB return db.get_records(hits);
Logic at the Edge via Boa
Extend the backend without recompiling. ApexKit embeds the Boa JavaScript engine to run server-side logic in a secure sandbox.
Event Hooks
Intercept DB operations (before_create, after_update) to enforce validation or trigger side-effects.
Cron Jobs
Built-in scheduler allows you to run maintenance scripts on a defined interval per tenant.
Custom Endpoints
Define manual routes accessible via /api/v1/run/{script_name}.
export default async function(e) {
const { amount, user_id } = e.data;
// 1. Database Lookup
const user = await $db.find_one('users', user_id);
// 2. Business Logic
if (amount > 1000 && !user.is_verified) {
throw new Error("Verification required.");
}
// 3. Side Effect
if (amount > 5000) {
await $http.post(process.env.SLACK_HOOK, {
text: "Big whale alert!"
});
}
return e.data;
}Dual-Engine Discovery
ApexKit combines Tantivy for lightning-fast full-text search (typo tolerance, faceting) with a Vector Store for semantic understanding.
- Instant Search: Memory-mapped inverted indexes for < 10ms response times.
- Semantic Search: "Find shoes for running" returns results even if the word "running" isn't present.
- Real-time Indexing: Data is searchable milliseconds after insertion via WAL hooks.
Ready to go extreme?
Download the binary, drop it on your server, and start building. No external dependencies required.