fix: update default model name and context size in config

document session registry

fix: respond when no tool used

stream only final responses

Add asynchronous message streaming

Update document upload logic and extend command execution timeout to 30 seconds

Remove unused EMBEDDING_MODEL_NAME from ChatSession initialization

Handle CancelledError during tool call follow-up

Add asynchronous VM execution and update prompt

Remove unused EMBEDDING_MODEL_NAME from configuration

Add Discord bot with file uploads

Remove server and API codebase, including FastAPI app, models, and router.

Add FastAPI server for LLM chat

Enhance system prompt

Handle multiple tool calls

Insert system prompt at runtime

Add per-user VM registry and update upload paths

Add document upload support and update prompt

Refactor execute_terminal function to set a fixed timeout and improve command validation

Add comment to EMBEDDING_MODEL_NAME indicating it is unused for now

Update model names and configuration parameters for improved performance

Add conversation reset command

Add persistent VM execution

Update docstring to specify execution environment as Linux terminal

Update docstring to clarify execution environment for terminal commands

Reduce timeout for terminal command execution from 30 to 3 seconds

Replace python execution tool with terminal

Update SYSTEM_PROMPT to include assistant name 'Starlette'

Remove add_two_numbers tool

Add 'system' role to Msg TypedDict for enhanced message handling

Add default system prompt and ensure it is used

Update NUM_CTX default value to 32000

Add configurable num_ctx option

Allow limited imports in execute_python

Add session-based conversation history

Update chat prompt and fix message type in ChatSession

Add user-based session tracking

fix: store tool calls as json and avoid duplicate messages

Add database support and persist chat history

Add initial implementation of chat application with required dependencies