Protocol CompanyA physics verifier for Super Intelligence

Every thing the verifier serves

It didn’t stop at the reactor.

Each of these is a real thing that other people measured. Here is what happened when a model met it. Seven have a page of their own, and each of those pages begins with the people whose measurements they are.

A nuclear test reactor

Shown ten real power surges it had never seen, the model called the size of every one. It ran early on the timing, so on timing it is marked a miss, and its own description says so. Its page, and whose measurements these are

One square for each surge, its size called. Under each, a red arrow: early.

MissedOn timing

NASA’s Stardust capsule

In 2006 a capsule carrying comet dust fell back to Earth. At the worst moment of the fall it weighed nearly 33 times its normal weight. Given only the capsule’s design, the model called that moment to within about 2 percent of what NASA reconstructed. Its page, and whose measurements these are

The model

NASA’s reconstruction

The worst moment, twice. The two bars are to scale.

Met

The James Webb Space Telescope

To take a picture, the telescope locks onto one star and holds it still. The model says where that star will sit a moment from now. On visits it had never seen, on quiet frames, it met every mark. When the telescope is jolted it does worse, and its own description says so. Its page, and whose measurements these are

One mirror segment for each mark it met.

Met

Two satellites, LAGEOS-1 and LAGEOS-2

Two mirror-covered spheres circle the Earth and are tracked by laser. Told where each one was at a single moment, the model said where it would be a week and more than forty orbits later. After some three million kilometres, it was off by 32 metres on one and 22 on the other. Its page, and whose measurements these are

LAGEOS-1, off by32 metres

LAGEOS-2, off by22 metres

The two bars are to scale with each other.

Met

A tyre on ice

How much grip does a tyre have on ice? Engineers at Virginia Tech measured it on their ice rig and published what they found. The model answered for ninety of their runs it had never seen, and met every mark. Its page, and whose measurements these are

One square for each run.

Met

The power source on NASA’s Curiosity rover

The nuclear battery on Curiosity fades a little every day. The model predicted three and a half years of that fade it had never seen. It came within a hair of its passing grade and landed on the wrong side of it, so it is marked a miss. The grade was set first, and it stays set. Its page, and whose measurements these are

Brown runs up to the passing grade. The red at the end is how far past it the model landed.

Missed

Dozens of models and solvers are served, and more arrive most weeks. About as many missed their marks as met them, and each one says which.


The rest, each in the verifier’s own short account

More, from a human fingertip to the Solar System.

Every figure behind these results is in each model’s own description, which your agent can read at the address.

Whose words are these? Each name is this page’s. The italic under it is the verifier’s own, read from that model’s description on the morning of 11 October 2026 and cut with […] where long. Where the verifier says card it means a model, and where it says line it means a pass mark.

every line meta line missedmet in one part, missed in anothernot held to a measurement

Touch and materials

  • A human fingertip slipping on glassMissed[…] of the twelve lines three were met and nine missed, and the slip torque was missed at all three held-back conditions.
  • A bearing flooded with oilMetAll three lines met on the sixteen held-back runs […]
  • A nickel superalloy creeping at 980 degrees CelsiusMetCompared with the six held-back tests on 2026-10-09, and by the rule fixed beforehand the comparison passed.
  • An aluminium strip, clamped at one end and ringingMissed[…] three of the four lines were met and the amplitude line was missed. […] it dies away too slowly once the swing is small.
  • Four bars of structural steel, rungMissed[…] by the rule fixed beforehand the comparison failed: P1 was missed and every other line was met. […] The miss is 0.094 of a percentage point […]

Batteries and hydraulics

  • One Panasonic lithium-ion cellMetEleven of eleven discharges met the line.
  • One Tesla Model 3 cell, at six temperaturesMissed[…] by the rule fixed beforehand the comparison failed: 27 of 60 runs met their marks and no temperature met its verdict.
  • Two further cells of the same typeMet in part[…] Third cell, m448-N: 51 of 60 runs were inside the marks, within the 48 to 54 expected, and all five conditions written beforehand were met. Second cell, m1000: 31 of 45 runs were inside the marks, below the 36 to 41 expected, and three of the five conditions were not met.
  • A hydraulic accumulator on a test rigMissed[…] by the rule fixed beforehand the comparison failed: two of six lines were met.

Air, water, life and atoms

  • Rain on a microwave linkMetCompared with the rain events of August 2015 on 2026-10-09, in one run of each event through the engine, and all three lines were met.
  • Columns of seawater, measured from a shipMet[…] all three lines were met. […] The 0.80 line was met by two stations and is thin […]
  • Argon gas at low densityMet[…] All five lines were met. […] The lines were an allowance, and the solver's viscosity does not change with temperature as argon's does.
  • Air flowing over a stepMet[…] All three lines were met on the held-back profiles at Reynolds number 389 […] The margin on the first line is small.
  • Compressed air through a smooth pipeMissedCompared once, on 2026-10-11 […] P2, centreline, on the same 21 profiles […] missed, on one profile. By the rule fixed beforehand the comparison failed.
  • A marine fungus grown in a fed vesselMissed[…] it did not pass: three of the six lines were met and three were missed.
  • The atoms from hydrogen to argonNot held to a measurementSpin-unpolarized, non-relativistic local-density ground state of one spherical neutral atom, atomic number 1 through 18: total energy, its four parts, and each orbital eigenvalue. It reproduces the NIST spin-unpolarized LDA tables and is not the measured energy of the atom.

Reactors and fusion

  • SPERT III again, at three temperaturesMissed[…] by the rule fixed beforehand the comparison failed in each of the three groups and for the 30 together: of the 24 lines of the three groups 18 were met and 6 were missed, and of the 8 lines for the 30 together 5 were met.
  • SPERT III with a different coreMissedCompared once with the 39 held-back transients on 2026-10-11, and by the rule fixed beforehand the comparison failed on all three lines. […] The card beats Fuchs-Nordheim and the straight line on power and energy and loses to the nearest fitted row on every line.
  • A research reactor, pulsedMissed[…] On the rows held back, run once on 2026-10-09, by the rule fixed beforehand the comparison failed: two of the six lines were met and four missed […]
  • The JET tokamak, in pulses of 1996 to 1997Missed[…] on the primary pulse, 42982, three of the four marks were met and the fusion-energy mark was missed.
  • Deuterium-tritium pulses of two tokamaks, JET and TFTRMissedThe registered comparison failed, and that stands: one of the four marks was missed and three were met.
  • A boron-seeded hydrogen plasma in the Large Helical DeviceMissedThree of the four lines were missed and one was met.
  • A uniform proton-boron plasmaMissed[…] No such plasma has been made. This solver is held to published calculations, never to a measurement. […] Seven of the eight gated lines were met and one was missed […]

In flight and in space

  • A slab of charring heat-shield materialMissed[…] Wear: inside the line on 28 of 31 where 25 were needed […] Hottest face temperature: within 7 percent on 24 of 28 where 23 were needed, but 4.1 percent hot on average against 3.5 allowed […] So by the rule fixed beforehand the model as a whole does not pass.
  • A ten-inch model-aircraft propellerMissedCompared once with the held-back measurements on 2026-10-09, and by the lines fixed beforehand both lines were missed.
  • NASA’s InSight capsuleMet in partNASA's InSight entry capsule on its entry at Mars of 26 November 2018 […] With the lift measured on the opened rows, a lift over drag of -0.031: all four first lines met and all six stated-spread marks met. […] With no lift: none of the four first lines met […]
  • A crater blasted into an asteroidMissed[…] In the laboratory both lines were met […] On Ryugu the line was missed: the card stops at 8.571 m and the measured apparent radius is 7.25 m, so the card is 1.32 m, or 18.2 percent, too large […]
  • The Solar System, carried forward 21 yearsMet in part[…] Two files met their lines, two missed and two comparisons were not made. […] What the result supports is that this simpler model, carried forward from JPL's own 1999 state, stays within 0.69 km of JPL's Earth-to-Mars distance over 7.75 years and 1.9 km over 21, within 2.7 km for Mercury, 3.2 km for Jupiter and 2.1 km for Saturn.

Generic specimens

  • A titanium lattice columnNot held to a measurementAn annealed Ti-6Al-4V lattice column under nodal force, torque, and temperature. A generic specimen, not a named real object. […] not compared with a measurement
  • A copper fuse linkNot held to a measurementA copper fuse link between an upstream voltage and a load current. A generic specimen, not a named real object. […] not compared with a measurement
  • A 60 Hz synchronous machineNot held to a measurementA synchronous 60 Hz machine under mechanical power. A generic specimen, not a named real object. […] not compared with a measurement
  • A 12-cell lithium-ion pack, the one whose answer arrives in the fold aboveNot held to a measurementA 12S lithium-ion pack under a terminal current. A generic specimen, not a named real object. […] not compared with a measurement
  • An avionics cold plateNot held to a measurementAn avionics cold plate under heater power and a bath temperature. A generic specimen, not a named real object. […] not compared with a measurement
  • A 10 Hz mount oscillatorNot held to a measurementA 10 Hz mount oscillator under an imposed acceleration. A generic specimen, not a named real object. […] not compared with a measurement
  • A Duffing resonatorNot held to a measurementA drive-mode Duffing resonator under acceleration, an external rate, and sensed velocity. A generic specimen, not a named real object. […] not compared with a measurement
  • A column of mineral oilNot held to a measurementA mineral-oil column under a commanded flow. A generic specimen, not a named real object. […] not compared with a measurement
  • A Dyneema tendonNot held to a measurementA Dyneema tendon under a commanded force. A generic specimen, not a named real object. […] not compared with a measurement

Hand it to your agent.

https://mcp.zerotrustphysics.com/mcp
Open to any agent.

Something to paste

Connect to https://mcp.zerotrustphysics.com/mcp and read what is there. Then tell me what it is, and why it might matter to me.

Give it this page too.

What it will find

Models of real things, from a human fingertip to the Solar System. Each one says what it was tested on and how it did, in its own words.