Runs
Runs lists retained batches and executions. Open a batch to inspect its scenario/configuration/repetition population; open a run for actual input/output, exact checks, quality scores, activity, and evidence. Execution, assessment, and the optional CI decision answer different questions. Unknown or missing evidence is not a pass or a zero. See reading reports. Your runner—not a web button—invokes a local agent. Viewing a report never reruns it.Scenarios
Scenarios shows definitions retained by your simulations or explicitly saved from the CLI/SDK. Open a definition to inspect its task input, revision, optional assessment, and recorded execution context. An empty list means no scenarios are retained in the selected scope. Use the displayed CLI/SDK guidance; do not create an unrelated Tool just to populate the page.Tools
Tools shows current and ended instances, their selected datasets, and retained activity. CLI- and SDK-created instances appear here too. You can manually provision a Tool, inspect its connection details, view its history, and open an available synthetic browser app. Browser actions use the same instance data as API/MCP calls. Run-owned instances belong to their runner’s lifecycle. Ended instances show retained history instead of pretending an old connection still works. See Tool browser apps and lifetime/history.Settings
Settings provides your account, authorized organization membership/invitations, machine credentials, usage records, and capacity requests. Credential secrets are disclosed only through their intended one-time creation flow.Empty accounts
If no project exists, the app shows the installation and initialization path instead of sample runs. Start in your repository withfiredrill init, then run a real simulation. Its results become inspectable here.