One database for every kind of data, written from scratch in Rust.
- All in one: caching, relational SQL, full-text search, documents, time series, analytics, graphs, vectors, geospatial, wide-column and objects, in one engine with one transaction system.
- Speaks what you already use: PostgreSQL, MySQL, SQLite, Redis, MongoDB, Elasticsearch and more, so applications switch without code changes.
- Messaging and queues built in: RabbitMQ (AMQP), MQTT, NATS, SQS and Kafka, in the same transactions as your data.
- Every query language: PostgreSQL, SQL Server (T-SQL), MySQL, Oracle and SQLite SQL, each as itself, plus Redis, MongoDB, Elasticsearch, CQL and graph queries, all on the same data.
- Works with your tools: drivers, ORMs, GUI clients and BI tools made for other databases.
- Migrating is restoring a backup: point EggDB at your SQL dumps, SQLite files or Redis and MongoDB backups, or let it follow your running database and switch over.
- About a second to download and a second to install: one small file, no dependencies.
- A client people love: a desktop, browser and command-line client designed for EggDB, with familiar layouts for people coming from other database clients.
- Drivers for every language: C#, Java, Node.js, Python, Go, Rust, PHP, Ruby and more, working with each language's ORMs.
- BI and pipelines built in: our own dashboards, reports and ETL, plus Spark, Iceberg and Trino compatibility for big data.
- AI-native: a built-in MCP server lets any AI agent discover, query and manage data safely.
- Single: one binary, which is both the server and an embeddable library.
- Lightweight and the fastest: measured against the best database for each workload, every release.
- Every scale, every way of scaling: from a single small file on a microcontroller to a cluster holding billions of billions of rows. Replicas, multi-primary, sharding, distributed SQL, serverless, tiered storage, edge copies and offline sync.
- Every data type: every type of PostgreSQL, SQL Server, MySQL, Oracle, SQLite, Redis, MongoDB and more, with each one's exact behavior.
- Everywhere: every OS, every CPU and every cloud.
Today: v0.2.0, an embedded key-value store in one file
Everything above is the goal; the first release people can use is smaller. v0.2.0 is an
embedded, crash-safe, transactional key-value store: one file, a Rust API, the eggdb command
and eggdb ui, a full-screen terminal UI, and eggdb sql runs the first parts of the PostgreSQL
dialect on the same file: tables, joins, GROUP BY with aggregates (docs/sql.md).
eggdb serve serves the file to psql and PostgreSQL drivers, with --mysql-listen to the
mysql client and MySQL drivers in MySQL's SQL, and with --tds-listen to Microsoft's
sqlcmd in Transact-SQL, on loopback only until EggDB has TLS, with one
writer at a time. eggdb sql --dialect sqlite and the embedded API run SQLite's SQL on the same
file, and --dialect oracle Oracle's (no Oracle wire protocol). eggdb mcp lets AI agents
such as Claude Code read the file over MCP, read-only, every call in an audit log
(docs/mcp.md). What is in it and what
is not: docs/releases/v0.2.0.md;
what is tested and what is not: docs/correctness.md; the plan:
docs/roadmap.md and docs/decisions.md.
Install: download one file
Each release on the releases page has one archive
per platform holding the eggdb binary, which needs nothing else (the Linux musl builds are
static and run on any distribution). On Linux on x86-64:
curl -fsSL https://github.com/eggspot/EggDB/releases/download/v0.2.0/eggdb-v0.2.0-x86_64-unknown-linux-musl.tar.gz | tar -xz
sudo mv eggdb-v0.2.0-x86_64-unknown-linux-musl/eggdb /usr/local/bin/
eggdb version
Other platforms swap the target: aarch64-unknown-linux-musl, aarch64-apple-darwin (Apple
silicon), x86_64-apple-darwin, and x86_64-pc-windows-msvc or aarch64-pc-windows-msvc as a
.zip. SHA256SUMS and gh attestation verify <archive> -R eggspot/EggDB check a download. To
build it instead: cargo build --release -p eggdb, and the binary is target/release/eggdb.
Use it from a shell
eggdb create app.db
eggdb put app.db user:1 Ada
eggdb get app.db user:1 # Ada
eggdb scan app.db --prefix user: # user:1 Ada
cat photo.jpg | eggdb put app.db photo # any bytes, up to 1 GiB
eggdb scan app.db --format json | jq .key # for scripts and agents
eggdb check app.db # "ok: every page passed its checksum …"
eggdb export app.db backup.jsonl # every entry as JSON lines; import loads it
eggdb ui app.db # browse and edit it full screen; q quits
eggdb sql shop.db --create -c "CREATE TABLE items (id integer PRIMARY KEY, name text)"
EGGDB_PASSWORD=choose-one eggdb serve shop.db --mysql-listen 3306 --tds-listen 1433 # PostgreSQL on :5432, MySQL on :3306, SQL Server on :1433
PGPASSWORD=choose-one psql -h 127.0.0.1 -U eggdb shop -c "SELECT * FROM items"
MYSQL_PWD=choose-one mysql -h 127.0.0.1 -P 3306 -u eggdb shop -e "SHOW TABLES"
SQLCMDPASSWORD=choose-one sqlcmd -S 127.0.0.1,1433 -U eggdb -d shop -Q "SELECT TOP 1 * FROM items"
Or work in one session, with history and tab completion:
$ eggdb shell shop.db
EggDB 0.2.0: shop.db, open read-write (1 table)
Type .help for help. SQL ends with ; and .quit or Ctrl-D quits.
shop=> insert into items values (1, 'egg'), (2, 'bread');
INSERT 0 2
shop=> select name from items order by id;
name
-------
egg
bread
(2 rows)
shop=> put greeting hello
shop=> get greeting
hello
shop=> .quit
Every command, the output formats, exit codes and the UI's keys: docs/cli.md.
Use it from Rust
[dependencies]
eggdb = { git = "https://github.com/eggspot/EggDB", tag = "v0.2.0" }
use eggdb::Database;
fn main() -> eggdb::Result<()> {
let db = Database::open("app.db")?;
db.write(|tx| {
tx.put("user:1", "Ada")?;
tx.put("user:2", "Grace")?;
Ok(())
})?;
let tx = db.begin_read()?;
for entry in tx.scan_prefix("user:") {
let (key, value) = entry?;
println!("{} = {}", String::from_utf8_lossy(&key), String::from_utf8_lossy(&value));
}
drop(tx);
db.close()
}
Transactions, scans, durability, errors and limits: docs/embedded.md.
Licensed under either the MIT License or the Apache License 2.0, at your option.
