mike-end-to-end

MIKE+ End-to-End (SOP)

The other mike-* skills describe one tool group each. This one says how they fit together, what must be true before a number or figure is reported, and when to stop.

1. Pick an operating mode (state it in the first reply)

If the goal only needs Mode A, do not touch mikeplus at all: it is faster and needs no license.

2. Standard run directory (create it, use it everywhere)

runs/<case>/
  00_model/        COPY of the whole model folder (.sqlite, .mupp, .dfs0, hotstart .res1d ...)
  01_inputs/       rain CSV/.dfs0 you produced, notes
  02_results/      the engine writes next to the model: 00_model/<name>_m1d - Result Files/*.res1d
  03_plots/        every PNG
  manifest.json    mike_manifest_write, written last

For Mode C use 00_model_baseline/ and 00_model_<variant>/: two independent copies.

3. Hard rules

1. Copy first, always the whole folder. mikeplus has no undo and the MIKE 1D engine writes result files (and a .log) next to the model; external files (rain .dfs0, hotstart, .mupp) are referenced relatively, so copying only the .sqlite breaks the run.
2. Gate on status, not ok. mike_run returns ok:true whenever the call returned. Usable = status in {ok, completed_with_warnings}; completed_with_errors / incomplete / unknown = do not use the results, surface errors, warnings, log_file.
3. Discover ids before using them. mike_model_info (setups, scenarios), mike_results_list (quantities, element ids), mike_get_values (current values). Never invent an element id, table or column.
4. Warm-up. Global peaks can sit at t≈0 (initial condition). Pass skip_hours to mike_results_summary / mike_results_compare / mike_results_flooding / mike_plot_profile and always report peak_time.
5. Variants are copies, not scenarios. mike_set_scenario only activates scenarios that already exist in the model (prepared in the GUI). Table edits made "inside" a scenario are not isolated (see mike-compare), so a what-if edit goes on a separate model copy.
6. License boundary. run / edit / scenario / import need MIKE+ + a checked-out license; results / plots / compare / rain / manifest do not. On a license failure ("license not available", demo mode, an edit that "did not persist", an import that "exceeds the demo limit") stop that branch and report it; do not retry in a loop, the seat is rate-limited and re-tries make it worse.
7. Evidence only. Report a number only from a tool return; report a figure only when the tool returned ok:true and png; say "not run" for anything skipped.