Simulation and Assertion Reference

Kessetsu compiles each source to canonical SPICE and runs the same typed analysis contract in native CLI and Web Hub. Native execution discovers a version-probed Ngspice executable; the browser uses the pinned Ngspice-derived worker runtime. Simulator console text is not the public API: datasets, measurements and assertions are versioned Rust Core objects.

Analyses

Each argument must be positive and dimensionally valid; DC sweep supports independent voltage/current sources. Multiple analysis kinds may coexist in one source. Operating-point, transient, and AC analyses are unique, while DC sweeps are unique per source; ambiguous repetitions fail semantic validation with KES-C009. The browser runs valid analyses independently in a cancellable Worker and Core combines them into kessetsu.simulation.v1.

Assertions

The form is assert metric(arguments) comparator threshold. Comparators are <, >, ==, <= and >=. Result states are:

• PASS: valid measurement satisfies the comparison.
• FAIL: valid measurement violates the comparison.
• ERROR: the metric could not be evaluated from the available typed data.
• SKIPPED: simulation did not complete, so the assertion was not evaluated.

Unsupported metric names and invalid argument shapes fail semantic validation with KES-C006 before a simulator starts. Strict inequalities are not loosened. Equality and inclusive comparisons use the versioned absolute/relative tolerance reported with kessetsu.assertion.v1.

For supervised agents and CI, the same assertions may live in an evaluator-owned .kessreq file and be supplied with kess test design.kess --requirements limits.kessreq. This mode forbids inline assertions in the design, records the exact requirement-file SHA-256, and optionally verifies --requirements-sha256. See the CLI reference for ownership and file behavior. Hash pinning detects changed limits; it is not a substitute for filesystem permissions controlled by the supervising process.

Measurements

Primitive reductions include value, min, max, absolute peak, average/avg and rms. Derived metrics include gain, low-pass bandwidth/cutoff, frequency, phase, output power, efficiency, THD, clipping and device dissipation. Exact formulas, windows, required datasets, current polarity and power sign are normative in engineering measurements.

CLI examples

kess simulate circuit.kess --format json
kess test circuit.kess --format json
Get-Content circuit.kess | kess test - --format json

Exit 3 denotes simulator/runtime failure; exit 4 denotes assertion FAIL, ERROR or SKIPPED. kess test also returns exit 4 and KES-T000 without launching the simulator when no assertions are defined; use kess simulate when verification is not intended. JSON stdout remains one parseable object. Use --include datasets,raw-log only for debugging large/raw data.

If a physical part assignment includes provided ratings, simulation/test JSON also includes part_stress (kessetsu.part-stress.v1) and its summary. Exceeding such a value does not change the command exit code: these records may be user-entered, conditions may be incomplete, and the electrical model may omit thermal/SOA behavior. Express required acceptance behavior with an assert or evaluator-owned .kessreq; use part stress to keep selection evidence visible.

Reproducibility boundary

Kessetsu fixes source, Core schema, generated netlist, model/package hashes, simulator identity and evaluation formulas. Floating-point samples may differ slightly across simulator builds and platforms, so parity gates compare engineering decisions with declared tolerances instead of treating recorded decimals as universal golden bytes.