Feature: Agents base uncovered lines
  As a developer
  I want Behave tests for agents/base.py
  So that uncovered lines are executed and verified

  Background:
    Given a basic test agent setup

  Scenario: Processing pipeline schedules the process wrapper
    When I patch asyncio create_task and send a message
    Then the processing pipeline should schedule the process wrapper task

  Scenario: Process wrapper handles tuple input and emits result
    When I process a tuple message through the wrapper
    Then the tuple message and context should be passed and emitted

  Scenario: Process wrapper handles non-tuple input with empty context
    When I process a non tuple message through the wrapper
    Then the non tuple message should be processed with empty context

  Scenario: Process wrapper wraps exceptions with ExecutionError
    When process_message raises an error inside the wrapper
    Then an ExecutionError should be raised with the agent name

  Scenario: process_message_sync runs coroutine on the event loop
    When I call process_message_sync on the agent
    Then the synchronous result should match the async process output

  Scenario: Dispose cleans up streams that expose dispose
    When I call dispose on an agent with disposable streams
    Then both input and output streams should be disposed if available

  Scenario: AgentWithMemory stores and recalls values with its lock
    When I remember a value and recall it later
    Then the remembered value should be returned from memory