Skip to content
Playground · teaching subset · baseline 0.8.6

Run Thirsty-Lang

This runs a real subset of the language in your browser — core and governed modules, requires/ensures contracts, and T.A.R.L. policies — via a portal-local lexer, parser, and tree-walking interpreter. It does not execute the Python package or produce cryptographic runtime proofs. Edit the program and the interpreter result, audit trace, and decision hash update live. For a policy block, ALLOW, DENY, or ESCALATE is a policy disposition, not a terminal Project-AI execution verdict.

program.thirsty
Interpreter result / policy disposition
ALLOWmodule hello · core

Execution completed without a governing denial.

Output (pour)
hello, governed world!
Audit trace
  1. 1·module hello: core
  2. 2·glass greet(name)
  3. 3·greet("governed world") → "hello, governed world!"
  4. 4·drink message = "hello, governed world!"
  5. 5·pourhello, governed world!
  6. 6interpreter outcome ALLOWExecution completed without a governing denial.
Audit artifact
{
  "language": "Thirsty-Lang",
  "version": "0.8.6",
  "module": "hello",
  "mode": "core",
  "verdict": "ALLOW",
  "reason": "Execution completed without a governing denial.",
  "output": [
    "hello, governed world!"
  ],
  "decision_hash": "0xbaa47b4da7fd6c7abaa47b4d"
}

Decision hash is deterministic — identical source yields an identical hash, making interpreter results replay-friendly. The JSON field remains verdict for compatibility; ESCALATE means held for adjudication.