PVsyst CLI MCP

Unofficial, local-only Model Context Protocol server and dependency-free Python adapter for PVsystCLI 8.0.6 on Windows. It wraps the vendor's installed CLI; it does not bundle PVsyst, projects, weather data, license material, or simulation output.

Requirements

• Windows, Python 3.10+, PVsystCLI 8.0.6 and a configured PVsyst workspace.

• A PVsystCLI license or an available evaluation quota. The GUI and CLI have separate licensing states; check pvsyst_license_info before running simulations.

• An existing .PRJ project and .VC* variant in the workspace. The CLI does not create projects.

The run-simulation and convert-meteo flags used here were checked against PVsystCLI.exe help <command> for 8.0.6. Newer CLI versions may differ. A compatible SFI is needed for numeric result columns; without one, the generated CSV can contain only dates.

Related MCP server: blender-control-mcp

Install

cd path\to\pvsyst-cli-mcp
py -m pip install -r requirements.txt
$env:PVSYST_CLI = 'C:\Program Files\PVsyst8.0.6\PVsystCLI.exe'
$env:PVSYST_WORKSPACE = 'C:\path\to\PVsyst8.0_Data'
py -m unittest discover -s tests -v

Configure a local stdio MCP client with absolute paths (substitute your real paths and Python executable):

{
  "mcpServers": {
    "pvsyst": {
      "command": "C:\\path\\to\\python.exe",
      "args": ["C:\\path\\to\\pvsyst-cli-mcp\\pvsyst_mcp_server.py"],
      "env": {
        "PVSYST_CLI": "C:\\Program Files\\PVsyst8.0.6\\PVsystCLI.exe",
        "PVSYST_WORKSPACE": "C:\\path\\to\\PVsyst8.0_Data"
      }
    }
  }
}

Restart the MCP client after editing its configuration. This server uses the mcp Python SDK 2.x and exposes seven typed tools:

MCP file tools accept filenames, not arbitrary filesystem paths. Stage weather CSV and MEF files under workspace/Meteo, and SIT under workspace/Sites. Results go under workspace/Results, SFI definitions under workspace/Models. The MCP server does not expose an unrestricted shell or CLI escape hatch. It rejects existing output files rather than overwriting them.

Example with Python

from pathlib import Path
from pvsyst_cli import PVsystCLI, build_sfi, summarize

cli = PVsystCLI(cli_path=r"C:\Program Files\PVsyst8.0.6\PVsystCLI.exe",
                workspace=r"C:\path\to\PVsyst8.0_Data")
print(cli.license_info())
print(cli.list_projects())

# Replace project and variant with ones present in your own workspace.
sfi = build_sfi(cli.workspace / "Models" / "energy.sfi", "energy")
result = cli.run_simulation("MY_PROJECT.PRJ", "VC0", sfi=sfi,
                            out_csv=cli.workspace / "Results" / "run01.csv")
print(result)
print(summarize(result["csv"], "E_Grid"))

sum in the summary is the sum of samples. For an hourly CSV whose E_Grid unit is kW, it corresponds to kWh; check units and time step before interpreting other outputs. SFI output columns and variables vary by version and project. Predefined groups use identifiers observed in the official 8.0.6 example (viGlobInc, viE_Grid, viPR, etc.); custom vi* identifiers are accepted without implying that all are supported.

Checks performed

On a local 8.0.6 installation, an existing demo project completed a full-year hourly run in roughly 8-11 seconds. Both the vendor's SFI and an SFI generated by this adapter yielded 8760 rows with numeric GlobInc, E_Grid, and PR columns. CSV parsing and license-status queries were exercised; the included tests use mocks and temporary files, so they do not consume simulation quota. Actual behavior of convert-meteo, PDF generation, or other PVsystCLI releases requires separate validation.

Official references: PVsystCLI product, command reference, simulation use cases.

License

The source code in this repository is licensed under MIT. PVsyst and PVsystCLI are separate proprietary products; this integration is not affiliated with or endorsed by PVsyst SA.