SpiceMCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| LTSPICE_EXE | No | Path to the LTspice executable. If not set, LTspice is auto-detected. | |
| SPICEMCP_PROJECT | Yes | Path to the circuit project directory where state is stored. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
SpiceMCP designs circuits with LTspice. Typical flow:
start_from_template (filters, amplifier, LDO, buck: sized for you; see list_templates), or start_optimization (your netlist with {PARAM} placeholders and .meas lines; it checks the spec is possible first)
run_optimization (the server searches; you do not loop per simulation)
finish_design (standard values; yield over tolerance, supply, temperature; part stress and the rating to buy)
generate_design_report Real parts: find_part, then start_from_template(parts={"opamp": NAME}). Built-in blocks need no .subckt: X1 in+ in- V+ V- out SPMCP_OA (op-amp) or SPMCP_CMP (comparator). Long calls return within a minute; if an answer says "still running", call the same tool again. query_run reads saved state at any time. Follow next_step. Never rebuild a past circuit from memory: use query_run(what="netlist").
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| check_setupA | Show the LTspice path and project folder in use. Call first if anything fails to simulate. |
| simulateB | Run one netlist now and return its .meas results. Nothing is saved. To sweep, set sweep_param ("R1", "{GAIN}" or "temp") and sweep_values. |
| check_feasibilityA | Check if a spec is physically possible before simulating. Free, no LTspice. Use mode="concept" without a netlist; tech holds SI values like vdd, cload, vin. |
| list_templatesA | List ready-made circuits (filters, amplifier, LDO, buck) and the spec each needs. Start here when the user describes a circuit instead of giving a netlist. |
| find_partB | Find a real part by part number or maker in the SPICE libraries on this computer (LTspice's, plus SPICEMCP_LIB_DIRS). Use the name in start_from_template parts. |
| start_from_templateB | Build a run from a ready-made circuit, size it by hand equations and simulate that start point. spec keys from list_templates (SI); parts: names from find_part. |
| start_optimizationB | Create a saved optimization run. netlist uses {NAME} for each param and a .meas line per metric. Refused if the spec is impossible. |
| run_optimizationB | Let the server search for |
| evaluate_candidateA | Simulate one exact set of param values inside a run (cached if repeated). Use to test your own idea; use run_optimization for searching. |
| query_runC | Read saved run data. Default what="status". "runs" lists runs; "details" = hand calc, warnings, alternatives. candidate/netlist/similar need candidate_id. |
| rollback_to_candidateB | Restore a past candidate's exact netlist from storage and return its file path. Use this instead of rebuilding an old circuit from memory. |
| finish_designB | Snap the best design to standard values, check yield over parts (R 1% C 5% L 20%), supply and temperature (templates: 5%, -40/25/85 C); return the parts list. |
| generate_design_reportB | Write a Markdown design report for a run's best design and return its path. Optional checks cost LTspice runs: dc_audit, pvt (supply = source name), monte_carlo=N. |
| render_schematicB | Draw a schematic as PNG or SVG. path: an LTspice .asc, or a SPICE netlist (.cir/.net/.sp), which is auto-placed, routed and checked first. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| style_resource | Colours and fonts used by SpiceMCP schematics and plots. |
TDQS
Scored across 14 tools
Most tools target distinct workflow stages (feasibility, template start, optimization, evaluation, querying, finishing, reporting, rendering). The only real ambiguity is among the simulation family—simulate vs evaluate_candidate vs run_optimization—but descriptions explicitly distinguish one-off/no-save from in-run/cached vs let-server-search. check_feasibility also overlaps slightly with start_optimization's refusal logic, though it's framed as a free pre-check.
Nearly all names are snake_case with clear verb_noun or verb_object structure (check_setup, generate_design_report, start_optimization, evaluate_candidate, rollback_to_candidate). The one deviation is the bare verb 'simulate' with no noun, and 'finish_design' vs 'start_optimization' use slightly different framing, but overall the convention is predictable.
14 tools sit comfortably in the well-scoped 3-15 range. Each tool earns its place across the circuit-design lifecycle (setup, feasibility, templates, parts, simulate, optimize, evaluate, query, rollback, finish, report, render) with little apparent redundancy.
The surface covers a full design lifecycle: environment check, feasibility, template discovery, part lookup, simulation, optimization run creation, candidate evaluation, run querying, rollback, standard-value finishing, reporting, and schematic rendering. No obvious dead ends remain for the stated circuit-design purpose.