Skip to content

mcp_core.tools

The MCP tool surface: each function decorated with @mcp_tool is registered with the FastMCP server at startup. Pydantic schemas in schemas.py validate inputs before any rendering happens.

mcp_core.tools.diagram_tools

mcp_core.tools.diagram_tools

MCP tools for diagram generation using the decorator pattern

list_diagram_types 

list_diagram_types() -> Dict[str, Any]

Return the uml://types payload as a structured dict.

Keys are diagram type names (e.g. class, mermaid, c4plantuml); each value has backend, description, and formats.

generate_uml_batch 

generate_uml_batch(
    items: List[Dict[str, Any]],
    output_dir: Optional[str] = None,
) -> Dict[str, Any]

Render multiple diagrams; failures are collected per index without stopping the batch.

generate_uml 

generate_uml(
    diagram_type: str,
    code: str,
    output_dir: Optional[str] = None,
    output_format: str = "svg",
    theme: Optional[str] = None,
    scale: float = 1.0,
) -> Dict[str, Any]

Generate a diagram using the specified diagram type.

Use when you need a diagram of any supported type. For complex diagrams the default prompt guides planning first (type, elements, relationships), then you call this tool with the final code.

Parameters:

Name Type Description Default
diagram_type str

Type of diagram (class, sequence, activity, mermaid, d2, etc.)

 required
code str

Diagram code in the syntax for the chosen type

 required
output_dir Optional[str]

Directory to save the image. Omit or None for URL, playground, and content_base64 only (no file write; use in serverless / read-only).

 None
output_format str

svg, png, pdf, jpeg, txt, or base64 (default: svg). See uml://formats per type.

 'svg'
theme Optional[str]

PlantUML theme for UML diagrams (e.g. cerulean)

 None
scale float

Scale factor for SVG only (default 1.0, min 0.1). Ignored for other formats.

 1.0

Returns:

Type Description
Dict[str, Any]

Dict with code, url, playground; local_path when output_dir set; content_base64 when not saving; or error.

validate_uml 

validate_uml(
    diagram_type: str,
    code: str,
    output_format: str = "svg",
    strict: bool = False,
) -> Dict[str, Any]

Check inputs and light structure without rendering.

Use to catch unsupported types, bad output_format for the type, empty code, or obvious PlantUML/Mermaid/D2 issues before calling generate_uml.

Parameters:

Name Type Description Default
diagram_type str

Same as generate_uml (see uml://types).

 required
code str

Diagram source text.

 required
output_format str

Intended output format (default svg); must be allowed for the type.

 'svg'
strict bool

When True, apply extra Mermaid/D2 checks (no extra PlantUML rules).

 False

Returns:

Type Description
Dict[str, Any]

Dict with valid (bool), errors, suggestions, backend, diagram_type, prepared_code,
Dict[str, Any]

optional corrected_code when only trivial wrapping changed the text, and strict.

register_diagram_tools 

register_diagram_tools(server: Any) -> List[str]

Register all diagram generation tools with the MCP server

Parameters:

Name Type Description Default
server Any

The MCP server instance

 required

Returns:

Type Description
List[str]

List of registered tool names

get_tool_info 

get_tool_info() -> Dict[str, Dict[str, Any]]

Get information about all registered tools

Returns:

Type Description
Dict[str, Dict[str, Any]]

Dictionary mapping tool names to their information

mcp_core.tools.schemas

mcp_core.tools.schemas

Pydantic schemas for MCP tool inputs and outputs.

GenerateUMLInput

Bases: BaseModel

Input model for generate_uml tool.

ValidateUMLInput

Bases: BaseModel

Input model for validate_uml tool (subset of generate_uml; no output_dir).

DiagramResult

Bases: BaseModel

Structured output for diagram generation tools.

mcp_core.tools.tool_decorator

mcp_core.tools.tool_decorator

Decorator system for MCP tools

mcp_tool 

mcp_tool(
    name: Optional[str] = None,
    description: Optional[str] = None,
    category: str = "default",
    required_params: Optional[List[str]] = None,
    example: Optional[str] = None,
    annotations: Optional[Dict[str, Any]] = None,
) -> Callable[[F], F]

Decorator for registering a function as an MCP tool.

Parameters:

Name Type Description Default
name Optional[str]

Tool name (defaults to function name if not provided)

 None
description Optional[str]

Tool description (defaults to function docstring if not provided)

 None
category str

Tool category for organization

 'default'
required_params Optional[List[str]]

List of required parameter names

 None
example Optional[str]

Example usage of the tool

 None
annotations Optional[Dict[str, Any]]

MCP tool hints (readOnlyHint, destructiveHint, idempotentHint, openWorldHint)

 None

Returns:

Type Description
Callable[[F], F]

Decorated function
Example

@mcp_tool(description="Generate a class diagram", category="uml") def generate_class_diagram(code: str, output_dir: str) -> Dict[str, Any]: # Implementation return {"code": code, "url": "..."}

register_tools_with_server 

register_tools_with_server(server: Any) -> List[str]

Register all decorated tools with the MCP server

Parameters:

Name Type Description Default
server Any

The MCP server instance

 required

Returns:

Type Description
List[str]

List of registered tool names

get_tool_registry 

get_tool_registry() -> Dict[str, Dict[str, Any]]

Get the registry of all tools registered with the decorator

Returns:

Type Description
Dict[str, Dict[str, Any]]

Dictionary of tool metadata

get_tool_categories 

get_tool_categories() -> Dict[str, List[str]]

Get tools organized by category

Returns:

Type Description
Dict[str, List[str]]

Dictionary mapping categories to tool names

clear_tool_registry 

clear_tool_registry()

Clear the tool registry (mainly for testing)