AGiXT is a dynamic AI Agent Automation Platform that seamlessly orchestrates instruction management and complex task execution across diverse AI providers. Combining adaptive memory, smart features, and a versatile plugin system, AGiXT delivers efficient and comprehensive AI solutions.
The current implementation of this is in extensions/agixt_actions.py and will also be changed to use the new task planning functions. It is currently called Create Task Chain and Create Smart Task Chain, which will be replaced by plan_task.
What will plan_task do, you might ask? It will plan to complete tasks! But as a bonus, it will also create a new command to execute the chain of events for the agent and enable it once it is created. The agent will prompt the user to use the command once the chain is created. You can choose to execute the chain manually as well.
This is useful for multi-step tasks that might require some planning or research.
Implemented Conversational Memories
File uploads in conversations may pollute the agent's over all intended training if added to the main training collections. This change will grant the agents a short term memory in conversations that extends past the conversation history to be referenced when files are uploaded or websites are browsed within a conversation. You will still be able to add to the main collections 0-10 from the training interface.
Main changes with this include:
collection_number is now a string. I did not rename it to collection_id to avoid breaking changes, but this will now reference either one of the predefined numbers or the UUID for the conversation_id.
Browsed links are now browsed by conversation as well.
Add task planning functions
The current implementation of this is in
extensions/agixt_actions.py
and will also be changed to use the new task planning functions. It is currently calledCreate Task Chain
andCreate Smart Task Chain
, which will be replaced byplan_task
.What will
plan_task
do, you might ask? It will plan to complete tasks! But as a bonus, it will also create a new command to execute the chain of events for the agent and enable it once it is created. The agent will prompt the user to use the command once the chain is created. You can choose to execute the chain manually as well.This is useful for multi-step tasks that might require some planning or research.
Implemented Conversational Memories
File uploads in conversations may pollute the agent's over all intended training if added to the main training collections. This change will grant the agents a
short term memory
in conversations that extends past the conversation history to be referenced when files are uploaded or websites are browsed within a conversation. You will still be able to add to the main collections 0-10 from the training interface.Main changes with this include:
collection_number
is now a string. I did not rename it tocollection_id
to avoid breaking changes, but this will now reference either one of the predefined numbers or theUUID
for theconversation_id
.Add Single Sign-On
See notes from here: https://github.com/DevXT-LLC/MagicalAuth/pull/5