Playground
Write intent. Run it. Read the proof.
This is what beyond prompt engineering looks like: intent you can execute. Run a decision against real inputs and simulate a lifecycle against real events, deterministically and with no AI, then compile the whole mission into diagnostics, docs, a graph, a test plan, and a proof.
Run it
Executable intent
A decision is a program: give it inputs and it decides, first matching rule wins, with a full trace. A lifecycle is a state machine: give it events and it walks them, rejecting anything illegal. No code is generated. The intent itself runs.
Change an input, try age 16, or add a bad event like submit after approval, then run again. The trace on the right is the full audit of how the intent decided.
Compile it
From intent to proof
Load a mission and run the deterministic compiler. You get diagnostics, generated docs, a contract graph, a test plan, and a .intent-proof.json, the same artifacts the CLI emits.
Inline autocomplete and hover are powered by the compiler. Press Ctrl+Space for suggestions; hover a semantic type or note lens for help.
Click Run Compiler to compile this intent.
Deterministic. No AI. You get diagnostics, docs, a graph, a test plan, and a proof artifact.