spectre

Spectre Skill

Run a .scs netlist locally or on a remote machine through SSH, then parse PSF results into Python dicts. Independent of VirtuosoClient — no GUI needed.

Before you start

1. virtuoso-bridge is a Python CLI — install it in a virtual environment with uv pip install -e virtuoso-bridge-lite.
2. virtuoso-bridge status — check connection, Spectre path, license
3. Check examples/02_spectre/ — use existing examples as a basis
4. spectre must be on PATH, or set VB_CADENCE_CSHRC (project or user .env) so the runner can source the Cadence environment. This applies to local and SSH execution.

Core pattern

from virtuoso_bridge.spectre.runner import SpectreSimulator, spectre_mode_args

sim = SpectreSimulator.from_env(
    spectre_args=spectre_mode_args("ax"),  # APS extended (recommended)
    work_dir="./output",
)
result = sim.run_simulation("my_netlist.scs", {})

if result.ok:
    vout = result.data["VOUT"]
else:
    print(result.errors)

With Verilog-A includes:

result = sim.run_simulation(
    "tb_adc.scs",
    {"include_files": ["adc.va", "dac.va"], "spectre_args": ["+aps"]},
)

include_files and per-run spectre_args have the same meaning in local and SSH mode. They stage include files and extend the command; they do not inject arbitrary circuit parameters. For local-only execution, use SpectreSimulator.local(...), or configure VB_REMOTE_HOST=localhost and use from_env().

Result object

Treat result.ok as the execution contract. A nonzero exit, explicit fatal Spectre output, netlist read-in error, or explicit convergence failure returns FAILURE/PARTIAL even if the raw directory contains incomplete files. Do not use a non-empty result.data as proof that the simulation succeeded.