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·module hello: core
- 2·glass greet(name)
- 3·greet("governed world") → "hello, governed world!"
- 4·drink message = "hello, governed world!"
- 5·pour — hello, governed world!
- 6⊢interpreter outcome ALLOW — Execution 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.