Suggestion: Investigate context optimisers

[Sophist-UK]

Could save you a shelf of token costs by removing context from a call that isn't needed.

[Sophist-UK]

Also, MCP servers and other infrastructure can reduce input context cost and improve focus both of which save costs.