01
Decentralized Service Discovery
Register instances, monitor health, and let gateways or consumers react through SSE-driven notify-then-pull updates.
Decentralized Control Plane
Service discovery without the heavy machinery.
Litemesh is a lightweight control plane for service discovery, distributed KV, health checks, delayed jobs, and zero-trust mTLS. It gives teams a practical middle ground between ad-hoc service registration and a full service mesh.
Built For
Registry, KV, mTLS, MCP
Cluster Model
Peer-to-peer gossip
Operational Style
Small footprint, fast setup
What Litemesh can do
Litemesh focuses on the building blocks teams usually need first: discover services, watch for topology changes, distribute configuration, coordinate simple control-plane actions, and secure east-west traffic.
02
Distributed KV and Coordination
Store config, namespace-scoped values, CAS updates, sessions, and lightweight lock-style coordination primitives.
03
Built-In Zero-Trust mTLS
Use Litemesh as a practical identity and certificate layer when you want encrypted service-to-service traffic without a full sidecar mesh.
04
Async Control-Plane Tasks
Submit delayed jobs with callback and retry behavior for reminders, expiration workflows, and lightweight orchestration.
05
MCP-Ready Infrastructure
Expose tools, resources, and bridge-based MCP integrations for AI agents that need live infrastructure context.
06
Small Operational Surface
Start with a compact Go binary, familiar HTTP APIs, and a workflow that feels closer to shipping a registry than operating a platform stack.
Why teams look at Litemesh
Litemesh is not trying to be everything. Its value comes from choosing a lighter, more focused layer for service identity, discovery, and config propagation.
Where it shines
Good fit when you want less platform weight
- You need service discovery and distributed config without standing up a larger control plane.
- You want real-time change signals for gateways or consumers.
- You want mTLS building blocks without committing to a full sidecar mesh.
- You want a practical control-plane base for AI or MCP-driven infrastructure workflows.
How to think about it
Not a full service mesh, not just a registry
- Smaller and lighter than a full mesh stack.
- Higher-level than rolling your own around a raw KV database.
- Best treated as a decentralized identity, discovery, and config foundation.
- Designed for teams that want useful primitives first, then deeper platform decisions later.
How it works
The core interaction model is intentionally simple: write state once, emit a lightweight signal, and let clients or peers re-pull the latest truth.
1
Register
Services or tools register with metadata, namespace, and optional health configuration.
2
Propagate
Cluster peers spread changes through gossip and targeted synchronization behavior.
3
Notify
Consumers receive lightweight SSE change signals instead of constantly polling for state.
4
Reconcile
Gateways, apps, or agents pull the latest snapshot and update their local view safely.
A lighter control plane for teams that want discovery, config, and identity
without dragging in an entire mesh program on day one.
Litemesh is strongest when it is treated as a sharp, focused layer. It can help
you move from ad-hoc service registration toward a more coherent control plane,
while keeping the operational model approachable.
Get started fast
Start from the repository, docs, and release artifacts. This page is intentionally simple so it can evolve into a proper project website later without locking you into a bigger frontend setup today.