live
production ready
built in Go
bitcask-inspired

High-Performance
Embeddable KV

The pure Go key-value store built for speed. sub-microsecond latency, thread-safe embeddability, atomic compaction, and robust data integrity—accessible via Unix sockets, HTTP proxies, or directly in your Go binary.

$curl -sSL https://raw.githubusercontent.com/AatirNadim/getMe/main/install.sh | bash
Automated installation for Linux & macOS (amd64/arm64)
~0ns
Write Latency
~0ns
Read Latency
0 SDKs
Official Clients
AGPLv3
Open Source
getme-bundle

# Start the full getMe stack

$ docker-compose up -d

[+] Running 4/4

✔ Container getme-store Started :8080

✔ Container http-proxy-go Started

✔ Container grafana Started :3000

✔ Container loki-alloy Started

# Set a key-value pair

$ go run . put greeting "hello world"

OK

# Retrieve the value

$ go run . get greeting

"hello world"

getMe
Use Cases

Why choose getMe?

Purpose-built for local performance. If you are building single-node systems, getMe is designed to outperform general-purpose databases.

📦
vs. SQLite

Embedded Go Apps

No external daemon required. Import the getMe engine directly into your Go binary and use it in-memory or on disk with zero operational overhead.

🚀
vs. Redis

High-Throughput Local Caching

Sub-microsecond latency and IPC sockets make it vastly superior to network-bound tools for local, single-node caching layers.

🌐
vs. LevelDB

Edge & IoT Computing

A tiny memory footprint, zero dependencies, and pure Go architecture make it perfect for resource-constrained edge devices and IoT gateways.

Developer Experience

Anything you need.

From single-key operations to batch ingests across multiple language SDKs — getMe's API is intuitive from day one.

# Basic CLI operations
$ go run . put mykey "hello world"
→ OK

$ go run . get mykey
→ "hello world"

$ go run . delete mykey
→ OK
live • store.log
Ecosystem

Anywhere you need it.Any way you build it.

Built for the community, with ❤️.

Docker

Single container with persistence

x86ARM64

Go SDK

Native client with connection pooling

v1.21+

Java SDK

Async client for JVM ecosystems

17+
file_type_typescript_official

TypeScript

Browser and Node.js support

ESM

Python

Sync and async clients

3.9+
📊

Observability

Grafana, Loki, Prometheus ready

OTEL
Internals

Engineered for
Speed & Simplicity

System Architecture
CLI / REPL
getme-cli
Go SDK
UDS (Unix Sockets)
JS/Py/Java SDK
UDS (Unix Sockets)
HTTP Proxy
http-proxy-go · Port 8080
Storage Engine
Log-structured Hash Table · Segments
HashTable
In-memory index
SegmentMgr
Append-only log
Compaction
Atomic swaps
MCP Server
Grafana Alloy
Loki
IPC Sockets
📝
Log-Structured Storage
Fast disk I/O via append-only writes to active segments. SegmentManager handles rapid serialization and persistence of key-value payloads.
⚡
In-Memory Hash Index
Lightning-fast single disk seek using an in-memory HashTable to locate exact Segment IDs and byte offsets. No scanning. No guessing.
🔄
Background Compaction
Automatic atomic swaps of stale, dirty segments for clean, compacted ones via compactedSegmentManager.go. Disk usage stays bounded forever.
📦
Buffered Batching
Dramatically reduces system calls during ingestion. High-volume bulk writes are aggregated into memory pools and flushed to disk as large contiguous chunks to avoid I/O bottlenecks.
🔌
Unix Domain Sockets
Blazing fast local Inter-Process Communication bypassing the TCP stack. Extremely low-overhead routing for SDKs and the CLI.
Benchmarks

Built and Benchmarked
to Scale.

Transparent performance and correctness. These baselines were captured on a Linux AMD64 environment powered by an AMD Ryzen 7 5800HS processor. Expect even higher throughput on more modern or dedicated server infrastructure.

0
ns/op
Single Put
0
ns/op
Single Get
0
ns/op
90% Read / 10% Write
0
ns/op
Delete Operation
go test -bench . ./server/tests/... — Benchmark Results
BenchmarkDelete
69 ns/op
BenchmarkGet
187 ns/op
BenchmarkReadWriteMixed_90_10
1,933 ns/op
BenchmarkReadWriteMixed_80_20
2,562 ns/op
BenchmarkPut
4,663 ns/op
Reproducible — run on your own hardware:
go test -bench . ./server/tests/...
Detailed continuous benchmarking reports from our CI:
View Full Reports
AI & Agents

Context at theSpeed of Thought.

Seamless Model Context Protocol (MCP) Integration.

MCP Architecture
Claude Desktop
Cursor
getme-mcp-server
Stdio Transport
getMe Server
Fast Local Key-Value Store
MCP RegistryPyPI Published
CI/CD Automated
Official MCP Registry
Listed in the official Model Context Protocol registry. Tools like Claude Desktop can seamlessly discover and interact with your local getMe storage.
PyPI Published
Available as getme-mcp-server on PyPI. Install it instantly via uvx, pipx or directly in your Python projects.
AI Coding Assistants
Empower Cursor, Windsurf, and other agents to query metrics, inspect keys, and manage storage natively during your development workflow.
⬡ AGPLv3 Open Source

Powered by
Open Source.

getMe is licensed under AGPLv3. Contributions are welcome ❤️!..from reporting bugs and improving documentation to optimizing the core storage engine..

CONTRIBUTING.md
AGPLv3 License
CI Benchmarks
github GitHub

Ready to dive deeper?

Explore the official documentation to learn about architecture details, CLI commands, SDK usage, and deployment strategies.