# Tasklist
Keeping track of active tasks & statuses
## DONE
- test fastmcp connectivity
- test openai & env keys
- datamodel & primitive database
- implement memory CRUD tools
- dockerize mcp server
- unittest tools
- unittest llm
- sanitize user inputs
- tool call logging
- simple web client interface
- dockerize web-client
- mcp usage analytics: tool calls, consumed tokens
## TODO
- deployment plan
- mcp server as container -> for persistent connection with llm api
- website frontend deployment
- cli client - no deployment
- demo scenario - auto replay question set - playwright
- refactor & optimize
- document design decision
- guardrail - sensitive content
- replace OpenAI client with more unified interface & adapter
## DISCONTINUED
- database adapter -> technically feasible & straightforward -> use json now for ease of manipulation & testing
- authentication -> token-based authentication or more robust protocol e.g., OAuth
- admin privileges -> restricted functions e.g., list user -> should be implemented after authentication & authorization
- model performance test -> build set of difficult test cases, e.g., hard semantic similarity questions, to find best-performing models