Document first
Schema-free documents stored as readable files, with indexes and a packed format when collections grow.
AxiDB is an embedded, file-based document database for PHP. Copy it into your project and start storing documents. No server. No Composer. No third-party dependencies. Just a database you can understand.
AxiDB is designed for projects where installing and maintaining a database server feels heavier than the problem itself. Put the folder in your project, point it at a data directory and use the API.
// copy axidb/ into your project require 'axidb/core/axidb.php'; $db = axidb(__DIR__ . '/datos'); $db->insert('clientes', ['nombre' => 'Ana Ruiz', 'ciudad' => 'Murcia']); foreach ($db->find('clientes')->where('ciudad', '=', 'Murcia')->get() as $cliente) { echo $cliente['nombre']; }
CRUD, queries, indexes, transactions, backups and more — while keeping storage close to the application.
Schema-free documents stored as readable files, with indexes and a packed format when collections grow.
where, orderBy, limit, offset, select and joins, plus AxiSQL for richer queries.
Atomic updates with recovery after interruption and process locking.
Full and incremental backups, restoration, JSON and CSV.
Vector search for meaning-based retrieval, with a measured 45 ms result on 10,000 documents in the project benchmark.
Agents get a controlled database view with permissions, traceability and a stop control.
A collection can be a directory of JSON documents you can open, inspect and repair manually. Larger collections can use packed storage.
There is no separate service to install or keep alive. Authentication, roles, network access and TLS remain the responsibility of the host application.
The repository includes security testing and documents the boundaries of the embedded model.
Collections can be encrypted at rest, with HMAC protection for index bucket names.
Writes are designed to survive interruptions, with recovery and process locking.
AxiSQL keeps literals separate from syntax and data paths are constrained.
AxiDB is embedded, not a server. Authentication, roles, network controls and TLS belong to the host application.
// project status version 0.8.0 php 8.1 · 8.2 · 8.3 · 8.4 platforms Linux · Windows tests > 3,000 checks security 560+ attacks rejected license Apache 2.0
The repository compares AxiDB with SQLite on 10,000 documents. AxiDB does not try to beat SQLite at raw write speed; its trade is simplicity and zero installation.
Measured on a Windows laptop with SSD. Reproduce with php bench/comparativa.php.
PHP 8.1+ with the built-in JSON extension. No third-party dependencies are required.
Designed to work even on shared hosting where there may be no command line or database server.
Read the 5-minute guide ↗Copy axidb/ into your projectrequire 'axidb/core/axidb.php';$db = axidb(__DIR__ . '/datos');$db->insert('clientes', [...]);AxiDB is deliberately clear about where it fits — and where another engine is the better choice.
Useful when you want document storage, simple deployment and a small operational footprint.
AxiDB is embedded, does not provide MVCC isolation, is not designed for tens of millions of documents, and does not claim to outperform SQLite.
When a project needs more, AxiDB adds capabilities without becoming a separate platform.
JOINs, aggregates, GROUP BY, date and text functions, subqueries, views and EXPLAIN.
An HTTP bridge with tokens and CORS exposes the database to web clients.
An ES module client without dependencies or a bundler.
Vector search for applications that need semantic retrieval.
Profiles such as core, docs and ai let the engine enforce the intended database mode.
Describe, inspect statistics and run health checks with actionable warnings.
The short answers to the questions developers usually ask first.
AxiDB is an embedded, file-based document database for PHP. It keeps storage close to the application without requiring a separate database server.
No. AxiDB is designed to run inside the PHP application, so a separate database server is not required.
No. AxiDB is designed to work without Composer and without third-party dependencies.
Yes. AxiDB includes vector search capabilities for applications that need semantic retrieval.
Yes. The source is available on GitHub and the project is released under the Apache 2.0 license.
Use it when an embedded document database, simple deployment and a small operational footprint matter more than maximum database-server throughput or very large-scale storage.
AxiDB is built by GestasAI for developers who want the database close to their application — without unnecessary infrastructure.