collectTools — gather decorated functions

Python · package toolnexus · SPEC §6

Not available in Python

JS/Java/C# expose an explicit collector; Python, Go and Elixir have no equivalent public symbol.

Sweep a module or class and collect every function marked as a tool.

What to use instead

Python has no reflection-based sweep that walks a module or class and pulls out every @tool-decorated function automatically — there is no registry a decorator writes into, and @tool (see tool) rebinds the decorated name to the resulting Tool in place, so the functions are already sitting wherever you defined them.

Build the same outcome explicitly instead: keep a plain list[Tool] (or dict[str, Tool]) next to where the @tool-decorated functions live, and pass it straight to create_toolkit(extra_tools=...):

from toolnexus import tool


@tool
def add(a: int, b: int) -> str:
    """Add two numbers."""
    return str(a + b)


@tool
def ping() -> str:
    """Health check."""
    return "pong"


# The "collection" step, done by hand — one list, right where the tools are defined.
MY_TOOLS = [add, ping]

If the functions genuinely live scattered across a module and a decorator-based registry would be convenient, inspect.getmembers over the module combined with isinstance(obj, Tool) gets you the same sweep JS’s collectTools does — @tool already leaves a Tool instance bound to the name, so filtering a module’s members by type is the collector, without toolnexus needing to ship one.

See also

• define_tool — Wrap a plain function with a name, description and schema — the shortest path from code you have to a tool the LLM can call.
• tool — Derive the schema from the function signature or annotation instead of writing it by hand.