KynetraDB developer ecosystem
Keep familiar query and model workflows while KynetraDB remains honest about the boundary. Two native HTTP previews and ten Postgres bridge starters share one versioned capability catalog.
Direct to /rest/v1
Prisma-shaped repository methods and a Drizzle-shaped query builder translate a bounded, validated subset to KynetraDB HTTP requests. They are not full upstream ORM drivers.
A Prisma-shaped repository for findMany, findUnique, create, update, delete, and upsert. It is not a Prisma driver adapter and does not run Prisma query-engine plans.
Each native HTTP mutation is one request. Interactive and nested Prisma transactions are unsupported.
A Drizzle-shaped select builder with typed condition nodes and a bridge pgTable schema. It is not a drizzle-orm database driver.
Native HTTP builder requests are independent. Drizzle PostgreSQL transactions apply only when using the bridge database directly.
const posts = createPrismaRepository({ baseUrl, apiKey, model: 'posts' })
const published = await posts.findMany({
where: { published: true, views: { gte: 10 } },
orderBy: { created_at: 'desc' },
take: 20,
})const db = createDrizzleClient({ baseUrl, apiKey })
const published = await db.select('id,title')
.from('posts')
.where(and(eq('published', true), gte('views', 10)))
.orderBy(desc('created_at'))
.limit(20)Direct PostgreSQL access
Schema mappings target the same kynetra_entities table used by KynetraDB Postgres Bridge. KynetraDB owns that table, its indexes, and helper functions.
A SQLAlchemy 2.x declarative mapping for bridge entities and JSON attributes.
Transactions cover the bridge PostgreSQL database only; they do not create a distributed KynetraDB Universal Log transaction.
A JPA entity mapping for the bridge table with JSON and array columns.
Transactions cover the bridge PostgreSQL database only; they do not create a distributed KynetraDB Universal Log transaction.
An EF Core entity and model configuration for Npgsql-backed bridge deployments.
Transactions cover the bridge PostgreSQL database only; they do not create a distributed KynetraDB Universal Log transaction.
An unmanaged Django model for reading and writing bridge entities without owning migrations.
Transactions cover the bridge PostgreSQL database only; they do not create a distributed KynetraDB Universal Log transaction.
A Rails model for the bridge table with ownership and validation boundaries.
Transactions cover the bridge PostgreSQL database only; they do not create a distributed KynetraDB Universal Log transaction.
A Sequelize model definition for JSONB attributes and float-array embeddings.
Transactions cover the bridge PostgreSQL database only; they do not create a distributed KynetraDB Universal Log transaction.
A TypeORM entity mapping for the bridge-owned table.
Transactions cover the bridge PostgreSQL database only; they do not create a distributed KynetraDB Universal Log transaction.
A Diesel table declaration and queryable bridge entity.
Transactions cover the bridge PostgreSQL database only; they do not create a distributed KynetraDB Universal Log transaction.
A GORM model using pgtype for JSONB and float-array bridge columns.
Transactions cover the bridge PostgreSQL database only; they do not create a distributed KynetraDB Universal Log transaction.
An Ecto schema for bridge entities with migration ownership disabled.
Transactions cover the bridge PostgreSQL database only; they do not create a distributed KynetraDB Universal Log transaction.
Migrate in · 14 sources
One migrator CLI reads from 14 systems and writes Collections and entities into KynetraDB. Sources split by how they reach your data: a file on disk, a database wire protocol, or a vendor HTTP API.
| Source | Subcommand | Reaches your data via | What it reads |
|---|---|---|---|
| Cloudflare D1 | d1 | SQLite file | A D1 export. Tables are enumerated from sqlite_master, skipping SQLite internals and Cloudflare _cf_* bookkeeping tables. |
| PocketBase | pocketbase | SQLite file | A pb_data/data.db file, using PocketBase’s _collections meta table. |
| SQLitenewest | sqlite | SQLite file | Any local SQLite database. One collection per table, one entity per row; sqlite_% internals and all views skipped. |
| MongoDB | mongo | Wire proto | All collections from a database. |
| PostgreSQL | postgres | Wire proto | Schema-aware: creates Collections, then streams rows. |
| Supabase | supabase | Wire proto | A Supabase or PostgreSQL database. |
| Algolia | algolia | HTTP API | An index via the Browse API. |
| Appwrite | appwrite | HTTP API | All databases, collections and documents from a project. |
| Convex | convex | HTTP API | All tables from a deployment via internal migration functions. |
| Meilisearch | meilisearch | HTTP API | All documents from an index. |
| Pinecone | pinecone | HTTP API | An index via the Data Plane REST API. |
| Typesense | typesense | HTTP API | All documents from a collection, as an NDJSON export. |
| NDJSON | ndjson | File | A per-tenant export produced by the admin export endpoint; verifies the integrity trailer before it writes. |
| Firebase | firebase | JSON export | A Firestore admin export JSON file. |
The newest source opens the file with SQLITE_OPEN_READ_ONLY andno CREATE flag. That is a structural guarantee rather than a policy: SQLite itself refuses the write, so a migration cannot mutate, vacuum, or fabricate your data — and it cannot quietly create an empty database if you typo the path, because a missing file makes the open fail instead of succeeding against nothing. Journal and WAL sidecar files are not created either.
It matters because the failure mode it removes is the expensive one: a migration tool that opens read-write can leave your source subtly different from how it found it, and you will not know until you try to roll back to it.
# One collection per table, one entity per row.
kynetra-migrate sqlite ./app.db
# Opened with SQLITE_OPEN_READ_ONLY and no CREATE:
# - cannot mutate, vacuum or rewrite the source
# - no -journal / -wal sidecars created
# - a missing path ERRORS instead of being
# created as an empty database
# - sqlite_% internals and all views skipped
# (tables only in this slice)Native HTTP mutations are individual requests. Direct ORM transactions stop at the bridge PostgreSQL database and do not span the Universal Log, files, caches, or sinks.
Application migration tools may own application tables. They must not drop, rename, or retype kynetra_entities or Kynetra helper functions.
BM25 and vector operations remain raw-query or KynetraDB API operations until each framework has a verified native extension.
Review template fingerprints and retrieval evidence inKynetra Foundry.