.. AUTO-GENERATED from cli_pipeline.md -- do not edit directly.

All experiments cited in the manuscript follow the same FHOPS CLI pipeline so readers can replay each stage without bespoke tooling:

1. **Validate or snapshot scenarios.** Run ``fhops validate <scenario.yaml>`` for each bundle (e.g., ``examples/tiny7`` and ``examples/med42``) and, when
   needed, snapshot synthetic tiers via ``fhops synth generate out/synthetic_<tier> --tier <tier> --seed <seed>`` (one call per tier, fixed RNG seeds).
2. **Benchmark solvers.** Invoke ``fhops bench suite`` with explicit ``--scenario`` and ``--out-dir`` arguments (for the manuscript:
   ``docs/softwarex/assets/data/benchmarks/<slug>/``) plus the desired heuristics (``--include-ils``, ``--include-tabu``) and iteration budgets.
3. **Run the tuning harness.** Launch ``python docs/softwarex/manuscript/scripts/run_tuner.py`` (which calls ``scripts/run_tuning_benchmarks.py``; wrapped by
   ``docs/softwarex/manuscript/scripts/generate_assets.sh``) to produce leaderboard/comparison tables in ``docs/softwarex/assets/data/tuning/``.
4. **Replay schedules.** Call ``fhops eval-playback`` on the best SA/ILS assignments (deterministic and stochastic modes) so utilisation and downtime metrics
   land under ``docs/softwarex/assets/data/playback/<scenario>/<solver>/<mode>/``.
5. **Estimate costs.** Execute ``fhops dataset estimate-cost`` for each machine bundle to capture owning/operating/repair totals in
   ``docs/softwarex/assets/data/costing/``.
6. **Synthetic scaling sweep.** Use ``python docs/softwarex/manuscript/scripts/run_synthetic_sweep.py`` to regenerate runtime-vs-size CSV/JSON + plots under
   ``docs/softwarex/assets/data/scaling/``.

``docs/softwarex/manuscript/scripts/generate_assets.sh`` runs all six steps with the published budgets and writes repository-relative paths into the assets. Run
through ``make manuscript-benchmarks`` (``run_manuscript_benchmarks.sh``), it appends the commit, asset mode, runtime, and an SHA-256 hash of the committed
asset files to ``docs/softwarex/assets/benchmark_runs.log``, a single provenance log for all artefacts; ``make verify-assets-hash`` recomputes the hash.
