Configuration#

PyFluent-MCP is configured primarily through FLUIDS_MCP_* environment variables. You can set these in your shell, MCP client configuration, or a .env file loaded by your client.

General settings#

Variable

Effect

FLUIDS_MCP_SETTINGS_JSON

External JSON file for overriding the bundled settings schema

FLUIDS_MCP_LOG_LEVEL

Server log level (default INFO)

FLUIDS_MCP_DISABLE_SESSION_LOGS

1 turns off session logs

FLUIDS_MCP_MAX_STEPS

Cap on deterministic tool-loop iterations (default 30)

FLUIDS_MCP_INTENT_GUARD

0 turns off the run_code crash-signature guard (default on)

Execution and retrieval behavior#

PyFluent-MCP focuses on deterministic Fluent tooling and MCP transport exposure.

The repository provides:

  • validated Fluent execution

  • schema retrieval and grounding

  • backend abstraction

  • deterministic tool execution

  • MCP transport support

Transport security#

TLS certificate verification is enabled by default for outbound retrieval and HTTP operations.

Variable

Purpose

FLUIDS_MCP_CA_BUNDLE

Path to a PEM CA bundle (also reads SSL_CERT_FILE / REQUESTS_CA_BUNDLE)

FLUIDS_MCP_VERIFY_TLS

Enable or disable TLS verification (default enabled)

FLUIDS_MCP_HTTP_TIMEOUT

Timeout for outbound HTTP operations

Server command-line tool options#

The ansys-fluent-mcp console script accepts:

You can also launch the server as a module:

python -m ansys.fluent.mcp

Flag

Description

--transport stdio|http

MCP transport (default stdio)

--host / --port

HTTP bind address (when --transport http)

--backend KIND

Default backend kind until connect is called (ships pyfluent)

--log-level

Log level (default INFO)

Next steps#