max_iteration check in base workflow agent seems misplaced.

[YutackPark]

The code:

llama_index/llama-index-core/llama_index/core/agent/workflow/multi_agent_workflow.py

Lines 425 to 439 in c7d061e

	async def parse_agent_output(
	self, ctx: Context, ev: AgentOutput
	) -> Union[StopEvent, AgentInput, ToolCall, None]:
	max_iterations = await ctx.store.get(
	"max_iterations", default=DEFAULT_MAX_ITERATIONS
	)
	num_iterations = await ctx.store.get("num_iterations", default=0)
	num_iterations += 1
	await ctx.store.set("num_iterations", num_iterations)
	if num_iterations >= max_iterations:
	raise WorkflowRuntimeError(
	f"Max iterations of {max_iterations} reached! Either something went wrong, or you can "
	"increase the max iterations with `.run(.., max_iterations=...)`"
	)

In parse_agent_output, the num_iteration increased by 1 and the max_iteration check happens.

This is misplaced. parse_agent_output is right after LLM generate texts and before tool call. The produced LLM outputs are discarded and not used, wasting resource. This kind of check should be placed right before the resource use (LLM call).