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.
Improvements to webseach to memory and activity logging
Added new website summarization function in Websearch.py called summarize_web_content which will take the content from a scraped website and turn it into chunks to ask AI to summarize to store into memory. This will improve memory quality and reduce garbage being stored in memories from web searches.
Added summarize_content bool to get_web_content function.
Improved Youtube caption scraping to summarize them.
Renamed browse_links_in_input to scrape_website
Modified the /api/agent/{agent_name}/learn/url endpoint to use the scrape_website with a depth of 3 when a URL is to be learned. This will help gain deeper insights on website it is scraping.
Moved get_tokens function to Defaults.py so that it can be accessed outside of Interactions.py without causing circular imports.
If a vision_provider is set for the agent and use_visual_browsing is set to true in agent settings, it will screenshot websites while browsing them for deeper insights into the website from the vision provider.
Added a new feature where the agent will add to the conversation log any time it does an activity such as browsing a website or executing a command. [ACTIVITY_START] and [ACTIVITY_END] will wrap all autonomous activities.
Removed AUTONOMOUS_EXECUTION concept from the code. The point of AGiXT was to be an autonomous agent and supporting it going against the nature of how it was built was not feasible with future plans.
Improvements to webseach to memory and activity logging
Websearch.py
calledsummarize_web_content
which will take the content from a scraped website and turn it into chunks to ask AI to summarize to store into memory. This will improve memory quality and reduce garbage being stored in memories from web searches.summarize_content
bool toget_web_content
function.browse_links_in_input
toscrape_website
/api/agent/{agent_name}/learn/url
endpoint to use thescrape_website
with a depth of 3 when a URL is to be learned. This will help gain deeper insights on website it is scraping.get_tokens
function toDefaults.py
so that it can be accessed outside ofInteractions.py
without causing circular imports.vision_provider
is set for the agent anduse_visual_browsing
is set totrue
in agent settings, it will screenshot websites while browsing them for deeper insights into the website from the vision provider.[ACTIVITY_START]
and[ACTIVITY_END]
will wrap all autonomous activities.AUTONOMOUS_EXECUTION
concept from the code. The point of AGiXT was to be an autonomous agent and supporting it going against the nature of how it was built was not feasible with future plans.