Natural language constraints missing in implementation

[SheridanM551]

Description:
In the paper and presentation, ORFS-Agent is described as being able to accept natural language constraints, e.g.:

"Optimize metric X as much as possible, while not allowing metrics A, B, C to deteriorate by more than 2%"

However, in the current implementation:

• No CLI arguments or config fields exist for specifying natural language constraints.
• No parsing or processing code for natural language input is found in optimize.py or related scripts.
• No example inputs are provided to demonstrate this feature.

Expected behavior:

• There should be a mechanism (e.g., CLI flag, config field, or input file) where users can provide natural language constraints.
• The code should parse these constraints and integrate them into the optimization loop.

Questions:

1. Is the natural language constraint feature implemented in this repo, or is it still experimental/incomplete?
2. If implemented, could you provide an example of how users can specify constraints?
3. If not implemented, do you plan to release this part of the code?

Thanks for clarifying!