Need more efficient methods for model tool calls

[andrewxhill]

The trading simulator MCP currently requires individual API calls for each price query, token information lookup, and trade execution. This creates unnecessary overhead and consumes a significant number of steps in agent workflows, limiting their reasoning capacity.

Problem

When running trading agents against the simulator:

• Agents frequently make 10-15 sequential calls to get_price and get_token_info for different tokens
• Each call consumes a step from the agent's limited step budget
• This repetitive pattern wastes steps that could be used for deeper analysis and decision-making
• Some agents nearly max out their allowed steps on basic data gathering

Proposed Solutions

Enhance the trading simulator MCP server to support batch operations:
Add array support to existing methods:

• get_price: Accept an array of token symbols/tickers and return prices for all in a single call
• get_token_info: Accept an array of token identifiers and return details for all in one call
• execute_trade: Optionally accept an array of trade instructions to execute multiple trades in one call
  Allow tool list setting:
  This may already be supported but i didn't see it in the docs. Like the recall mcp server, i want the ability to not enable some tools when i start the server. they waste context space and i don't want my agents calling them (e.g., update profile, get leaderboard). if i could just list the tools i want enabled as an env var, that would be ahndy.