Today we’re excited to introduce sHEL — a volatile, literal-safe, automation-friendly data substrate that brings the reliability of a turtle shell to your data pipelines.

The Problem

Modern data interchange is fragile. JSON, YAML, TOML — they all suffer from the same fundamental issues:

  • Injection vulnerabilities — unescaped strings open the door to injection attacks
  • Parsing ambiguity — subtle syntax differences break parsers across implementations
  • Tooling fragmentation — every format needs its own toolchain

Our Solution

sHEL takes a different approach. Inspired by the Unix philosophy, sHEL treats data as a stream of literal-safe tokens that flow through composable pipelines.

Key Design Principles

  1. Literal-Safe by Default — Data is always treated as literals. No escaping, no injection.
  2. Volatile Operations — Fast, in-memory processing with optional persistence.
  3. Automation-First — Structured output designed for machine consumption.

Quick Example

Here’s what sHEL looks like in practice:

# Parse a log file and extract structured data
cat app.log | shel parse --format json | jq '.level == "ERROR"'

# The output is always predictable, always safe
{
  "timestamp": "2025-08-15T09:00:00Z",
  "level": "ERROR",
  "message": "Connection timeout",
  "literal": true
}

What’s Next

This is just the beginning. We’re building:

  • A rich plugin ecosystem
  • IDE integrations
  • Visual pipeline builders
  • Enterprise features

Get started today and let us know what you think!