Runtime Mode

Tools for managing runtime mode and inspecting MCP client capabilities.

`get_runtime_mode`

Check if runtime mode is enabled and what capabilities are available.

Returns

Type: RuntimeModeStatus

Status including enabled state, registered tools, and availability.

Example

status = get_runtime_mode()
# Returns: RuntimeModeStatus(
#   enabled=False,
#   tools_registered=[],
#   docker_available=True,
#   oot_available=True
# )

`enable_runtime_mode`

Enable runtime mode, registering all runtime control tools.

This adds tools for:

XML-RPC connection and variable control
ControlPort/Thrift for performance monitoring
Docker container lifecycle (if Docker available)
OOT module installation (if Docker available)

Returns

Type: RuntimeModeStatus

Updated status with newly registered tools.

Example

status = enable_runtime_mode()
# Returns: RuntimeModeStatus(
#   enabled=True,
#   tools_registered=[
#     "launch_flowgraph", "list_containers", "stop_flowgraph",
#     "connect", "list_variables", "get_variable", "set_variable",
#     ...
#   ],
#   docker_available=True,
#   oot_available=True
# )

Notes

Calling when already enabled is a no-op (returns current status)
If Docker is unavailable, container-related tools are not registered
If Docker is unavailable, OOT tools are not registered

`disable_runtime_mode`

Disable runtime mode, removing runtime tools to reduce context.

Use this when you’re done with runtime operations and want to reduce the tool list for flowgraph design work.

Returns

Type: RuntimeModeStatus

Updated status showing disabled state.

Example

status = disable_runtime_mode()
# Returns: RuntimeModeStatus(
#   enabled=False,
#   tools_registered=[],
#   docker_available=True,
#   oot_available=True
# )

`get_client_capabilities`

Get the connected MCP client’s capabilities.

Useful for debugging MCP connections and understanding what features the client supports.

Returns

Type: ClientCapabilities

Client information and capability flags.

Example

caps = get_client_capabilities()
# Returns: ClientCapabilities(
#   client_name="claude-code",
#   client_version="2.1.15",
#   protocol_version="2024-11-05",
#   roots=RootsCapability(supported=True, list_changed=True),
#   sampling=SamplingCapability(supported=True, tools=True, context=False),
#   elicitation=ElicitationCapability(supported=False, form=False, url=False),
#   raw_capabilities={"roots": {"listChanged": True}, ...},
#   experimental={}
# )

Capability Explanations

Capability	Description
`roots`	Client exposes workspace directories
`sampling`	Server can request LLM completions
`elicitation`	Server can prompt user for input

`list_client_roots`

List the root directories advertised by the MCP client.

Roots represent project directories or workspaces the client wants the server to be aware of.

Returns

Type: list[ClientRoot]

List of root directories.

Example

roots = list_client_roots()
# Returns: [
#   ClientRoot(uri="file:///home/user/project", name="project"),
#   ClientRoot(uri="file:///home/user/gr-mcp", name="gr-mcp"),
# ]

Notes

Returns empty list if roots capability is not supported
Typically includes the current working directory