CTRLRun Execution Boundary

Tools

• ctrlrun_execution_boundary_run_scenarios Run the five scenarios shipped as `ctrlrun demo`, in process, with no network. This calls the library's own `run_demo`, not a re-implementation, so what you read here is exactly what `pip install ctrlrun && ctrlrun demo` prints. Each scenario raises if the action it expects to be blocked goes through, so a green run is evidence and not decoration. Returns: The demo's output, verbatim.
• ctrlrun_execution_boundary_run_walkthrough Run the console's whole six-step sequence against one fresh kernel, in one call. This is the stateless companion to the interactive console: it builds its own `Control`, store and fake remote, drives every step in order, and returns the transcript. Use it when you want the full sequence without holding a session — an MCP tool call, for instance, gets a fresh state each time, so a step-by-step API would silently lose the very history that makes `DuplicateEffect` and the consumed-approval replay fire at all. Returns: The transcript of all ten presses, with the remote's call list after each.

No tool matches.

As the server's own listing gives them. We never call a tool.