auto_func

auto_func turns a function into a model‑backed interface:

Signature → input structure
Docstring → instruction
Return annotation → output schema

Example

python

from agently import Agently

agent = Agently.create_agent()

@agent.auto_func
def calculate(formula: str) -> int:
  """
  Return result of {formula}.
  MUST USE TOOLS TO ENSURE THE ANSWER IS ACTUAL NO MATTER WHAT.
  """
  ...

result = calculate("3333+6666=?")
print(result)

Async functions

python

@agent.auto_func
async def plan(task: str) -> dict:
  """Return a plan for {task}."""
  ...

result = await plan("Launch a feature")

Notes

Generator functions are not supported
Return annotation is used as the output schema