Skip to Content
BlogIntroducing LODE

Introducing LODE

March 10, 2026

LODE is a deterministic binary serialization language designed for encoding structured data into compact, verifiable payloads. Today we are releasing v0.1.2 — the first stable version of the language, compiler, and toolchain.

Why LODE?

In distributed systems, data passes through many hands. Model configurations move across registries. Prompt contracts wrap inference requests. Configuration manifests get cached, replicated, and verified at every hop.

Existing serialization formats — JSON, Protocol Buffers, MessagePack — all share the same gap: they don’t guarantee that two identical objects produce the same bytes. Key ordering varies. Encoding choices differ. The result is that you can’t hash a payload and expect the hash to be stable.

LODE closes that gap. Every encoding is deterministic. Every fingerprint is reproducible. Two nodes with the same data will always agree on the hash.

What Ships in v0.1.2

  • Schema definition.lode files describe typed structures with field IDs, names, types, and required flags
  • Vein binary format — Compact wire encoding with a 4-byte header and typed field entries
  • Four primitive typesstring, bool, string[], map<string,string>
  • lodec CLI — Compile, decode, fingerprint, verify, and diff from the command line
  • Programmatic APIcompile(), decode(), fingerprint(), verify(), loadLode()
  • Forward compatibility — Decoders skip unknown fields, so v1 consumers can read v2 payloads

Try It

npm install @stacknet/lode lodec compile config.json -s schema.lode -o config.vein

Or use the Playground to experiment in your browser.

What’s Next

We’re working on integer and float types, streaming encoders for large payloads, and schema-to-TypeScript code generation. Follow the Version History for updates.

Last updated on