ONE LANGUAGE.
THREE LANES.

observability · analytics · secure-by-default

Lateralus is a scripting language with an unusual standard library. Fifty-plus modules, written in pure Lateralus, covering the exact primitives you need to build observability pipelines, analytical data tools, and cryptographic network services — without reaching for a C library, a Rust crate, or a Python wheel.

If the shape of your work is "parse a wire format, transform a stream, emit a secure artifact," this language fits your hand.

◉ OBSERVABILITY LANE

The Lua of telemetry pipelines. Ingest any wire format, transform with pipelines, re-emit to any downstream.
Alternative to: Vector, Fluent Bit, Logstash, Benthos — without a YAML DSL.

◉ ANALYTICS LANE

The columnar-native scripting language. Read Parquet, stream Arrow, compose pipelines.
Alternative to: Python-on-Arrow, DuckDB UDFs, Polars scripts — compiled to native C99.

◉ SECURE-BY-DEFAULT LANE

TLS-ish primitives in the stdlib. No OpenSSL linkage, no "just trust the wheel."
Alternative to: OpenSSL bindings, libsodium bindings, cryptography.io — all audit-readable in pure Lateralus.

◉ WHAT MAKES THIS DIFFERENT

No major scripting language ships all three lanes in its standard library. Python has hashlib but no Parquet. Lua has nothing. Ruby needs a dozen gems. Go has TLS but not Arrow. Rust has Arrow and TLS, but it's not a scripting language — compile-edit-test cycles are minutes, not seconds.

Lateralus compiles to a tree-walking interpreter for iteration, to C99 with gcc -O2 for production, and the stdlib is written in Lateralus itself — every byte is auditable, every algorithm is a readable textbook. No opaque binary shipping with your binary.

◉ ONE EXAMPLE, THREE LANES

// A log-shipper that reads Prometheus scrape lines, filters with
// Aho-Corasick for security keywords, batches into Parquet on disk,
// and encrypts the daily rollup with ChaCha20-Poly1305.

import prometheus
import aho_corasick
import parquet
import chacha20
import poly1305
import hkdf
import x25519

let scrape = http_get("http://prom:9090/metrics")
let metrics = prometheus_parse(scrape)

let rules = aho_corasick_build(["error", "panic", "denied", "timeout"])
let alerts = filter(metrics, fn(m) {
    aho_corasick_contains_any(rules, m.labels["message"])
})

let body = parquet_encode_rows(alerts, ["name", "value", "timestamp"])
let footer = parquet_footer(schema_for(alerts))
let file_bytes = parquet_assemble_file(body, footer)

let shared = x25519_scalar_mult(my_sk, peer_pk)
let key = hkdf_derive_key(salt, shared, bytes_of("rollup-2026-04-24"), 32)
let sealed = chacha20_xor(key, nonce, 1, file_bytes)
let tag = poly1305_mac(key, sealed)

write_file("rollup.parq.enc", sealed + tag)

Every function call above resolves to a pure-Lateralus stdlib module. Zero external dependencies. One binary.

TRY IN PLAYGROUND READ THE DEEP-DIVES STAR ON GITHUB