Changes in version 0.5.0                        

  - Adding the per-station retry features to the Workflow class
  - Adding the fallback handlers to the Workflow class

                 Changes in version 0.4.0 (2026-05-27)                  

Implementing the Workflow class which allows one to have full control over a workflow of agents

Adding new methods:

  - share_context_with()

                 Changes in version 0.3.0 (2026-01-12)                  

Adding new methods to work with tools:

  - register_tools()
  - list_tools()
  - remove_tools()
  - clear_tools()
  - generate_and_register_tool()

Adding the agents_dialog() methods that allows 2 agents to interact with each other in order to come up with a better outcome.

                 Changes in version 0.2.2 (2025-12-04)                  

  - Fixing syncing issues between mini007 and ellmer

Adding the following new methods:

  - validate_response()
  - clone_agent()

Deletion

  - Deleting the visualize_plan() method as the DiagrammeR package has
    many dependencies.
  - Deleting the add_message method for compatibility with ellmer Turns

Dependency

  - Adding ellmer as an import dependency.

                 Changes in version 0.2.1 (2025-11-03)                  

  - Fixing bug in the generate_execute_r_code() method.

                 Changes in version 0.2.0 (2025-10-29)                  

  - Setting messages as an active R6 field. It is now possible to modify
    the messages that will be used by the LLM object through a list
    object. The list object will be automatically converted to the
    corresponding ellmer Turns.

Adding the following new methods:

  - keep_last_n_messages()
  - update_instruction()
  - clear_and_summarise_messages()
  - judge_and_choose_best_response()
  - set_budget()
  - export_messages_history() and load_messages_history()
  - reset_conversation_history()
  - add_message()
  - generate_execute_r_code()
  - visualize_plan()
  - set_budget_policy()

Adding the following new parameters:

  - Adding the force_regenerate_plan boolean parameter to the invoke
    method of the LeadAgent, this will allow taking into account if the
    LeadAgent has already generated a plan or not. If a plan is
    detected, no need to generate the plan from scratch, except if the
    user set the force_generate_plan to TRUE.

Deleting the following method:

  - delegate_prompt(), not needed anymore as the generate_plan() method
    has the same behavior.

                 Changes in version 0.1.0 (2025-09-05)                  

  - Initial CRAN submission.