fastmcp.tools.base
Functions
default_serializer
Classes
ToolResult
Methods:
from_mcp_result
to_mcp_result
InputRequiredToolResult
The full result of a single multi-round-trip leg (SEP-2322).
The protocol is stateless: each MRTR leg is a complete request→response
cycle. When a guard tool returns an InputRequiredResult from its body to
ask the client for input, that ask is the legitimate result of this tool
call — not a pause, not an error, not a third control-flow outcome. FastMCP
wraps it in this ToolResult subclass so it flows through the middleware
chain as an ordinary return value: call_next(...) returns it, default
middleware completes normally on the leg, and middleware authors can
identify an ask with a simple isinstance(result, InputRequiredToolResult)
check.
Invariant: the wrapped InputRequiredResult is never serialized as tool
content. content is always empty; the wire handler (_on_call_tool)
reads .input_required and returns it to the runner as the
input_required result. Do not read .content / .structured_content on
this subclass — they carry nothing.
Tool
Internal tool registration info.
Methods:
to_mcp_tool
from_function
run
run() can EITHER return a list of ContentBlocks, or a tuple of
(list of ContentBlocks, dict of structured output).
A tool that requests client input (SEP-2322 multi-round-trip) does so by
returning an InputRequiredResult from its body; the run machinery wraps
that in an InputRequiredToolResult — a ToolResult subclass — so it
stays inside the declared ToolResult result type and flows through the
middleware chain as an ordinary result (see FunctionTool.run).