What the analyses model, and what they do not
Every solver on this site, from a section polar to a trimmed aircraft, assumes the flow is steady, rigid and fully submerged. The list of what that leaves out is the same list the MCP server gives every connected assistant in its instructions, read from one constant, so what an agent is told and what this page says cannot differ. The three words first, then the list.
Steady
Nothing changes with time. A result is for one speed and one angle, held; the solvers have no time axis, so there is no gust response, no flutter and no transient at the start of a manoeuvre. A speed sweep is a row of separate steady solutions, not a motion.
Rigid
The geometry analysed is the geometry you drew. Loads are reported, but they do not bend or twist anything, and the section keeps its shape at every speed.
Fully submerged
The fluid is unbounded. Air is unbounded in practice; water is not, and a hydrofoil is solved as if it were deep in it. The density and viscosity you pick set the Reynolds number, and that is the whole of the fluid's part in it.
What is not modelled
The free surface
A hydrofoil is solved as if the water went on forever above it. Nothing models the surface: not the loss of lift as a foil nears it, not air drawn down a strut (ventilation), not a tip breaking through.
Cavitation
The pressure on the suction side is never compared with vapour pressure, so a section that would cavitate at speed reports the lift and drag it would have if the water held together.
Unsteady motion
Every result is for a fixed speed and angle. Pumping, flapping, gusts and the start of a turn are unsteady, and the solvers have no time axis.
Flex and aeroelastic twist
The wing is rigid. A real wing bends under load and twists as it bends; the analysis sees the geometry you drew, not the one that flies.
Roll and yaw
Everything is in the pitch plane. Trim balances lift, weight, drag and thrust about one axis; there is no sideslip, no roll rate and no lateral stability figure.
Joined and box wings
A wing is one surface from root to tip. Trim places two of them about a CG, but neither solver handles surfaces that meet, close a loop or sit directly in line with each other.
Structure and manufacturing
Loads are reported; nothing says whether the spar carries them. There is no layup, no mass estimate and no check that the shape can be built.
Junctions
Where a wing meets a fuselage, mast or pod, the flow interferes and drag rises. Each part is solved on its own and the sum has no junction term.
The board and the rider
Trim takes the rider as a mass at a height and a board as a drag area. Neither lifts, neither moves, and the board never touches the water.
The trim page's own list
Trim adds non-lifting parts to the balance and says, per craft, what each template leaves out. The hydrofoil template's caption:
Both surfaces are solved together on a lifting line with the wake following the flow, so the stabiliser sees the front wing's downwash. Section drag is viscous; the mast is a 2D section at its own Reynolds number. Cavitation, the free surface and the mast's own lift are not modelled.
The RC template's:
Wing and tail solved together on a lifting line with the wake following the flow. The CG sits on the chord plane and the thrust line passes through it, so the balance is the classic one: the tail's arm against the wing's pitching moment. No fuselage is modelled.
The full-scale template's:
With the CG below the wing, the lift's forward component at positive α is a nose-down term, the opposite sign to the hydrofoil. The tail sits below the wing, in the region of higher downwash at cruise. The fuselage enters as drag at the CG height; its destabilising moment is not modelled.
The page's own notes add that a mast or strut enters as drag at a height and a board, pod or fuselage as a drag area, so each changes the thrust the balance demands and the moment about the CG but not the pitch stiffness, and that the free surface, cavitation and a mast's own side force are not modelled. The static margin it reports is the wing-plus-tail figure; a fuselage's destabilising moment is not in it.
What an agent is told
A connected assistant carries the list above in the server's instructions, with one standing rule: mention a limit in one sentence only when the question or the advice turns on it, and otherwise leave it out. If your assistant volunteers that the tools do not model something, that is the rule working. MCP tools has the tools; For agents has the rest.