master
prompt consistency and updates\nadded new performance evaluations
added langchain sequential chain for task to agents generation
memory exclusivity for local memory managment now working. fixed some small issues.
slugified filenames for *.yaml agents, refactored expert/multigpt modules into separate package
fixed issue where gpt-4 was hardcoded into the lmql queries and made it impossible to run multi-gpt without a valid gpt-4 key.
refactored code, implemented gpt-based smart selection to intelligently select next participant in discussion, fixed some bugs, added chat only mode, tweaked agent behavior to be less collaborative, added behavioral traits for each agent
cleaned up code
fixed merges issues.
Merge branch 'experimental-discord-integration'