No user-submitted related servers found.
Scored across 37 tools
Most tools occupy a clearly distinct slot in a real simulation workflow (case lifecycle, meshing, solving, results, sweeps, surrogates, topology optimisation, print screening). A few boundaries are fuzzy: instant_estimate vs surrogate_predict (both 'instant answer with error bar'), check_convergence vs run_sweep with study="convergence", and describe_case vs validate_case all require reading the descriptions closely. The thorough descriptions largely resolve these, but misselection is still plausible.
The dominant pattern is verb_noun snake_case (create_case, import_geometry, run_simulation, probe_results, build_surrogate), applied consistently across action tools. A handful of resource-first names (job_status, surrogate_status, genesis_result, environment_check) and one verbless name (instant_estimate) break the pattern, though they form a readable sub-convention for status/artifact tools.
37 tools is well past the comfortable range, but the surface genuinely spans many sub-domains: case CRUD, CAD import, meshing, solving, jobs, results inspection, verification, exports, parameter sweeps, surrogates, topology optimisation, and a standards-based estimator. Nothing is obviously trivial, yet several clusters (four surrogate tools, four result-inspection tools, two convergence paths) make the set heavier than it needs to be.
Case lifecycle is essentially complete (create/import/open/clone/update/undo/delete/export/list/describe/validate), and the solve-inspect-verify-export loop is fully covered, including background jobs and stopping them. The main gap is run management: get_results accepts 'the given run' but no tool lists a case's runs or deletes a single run, so agents working with multiple runs per case hit a dead end. There is also no standalone way to discover valid material presets or parameter paths other than describe_case(detail="full").