Improve Trading Assistant prompt clarity and agent naming

[ManuelKugelmann]

Summary

Updated the Trading Assistant's live chat prompt and agent configurations to improve clarity, consistency, and user experience. Changes focus on standardizing agent display names, removing internal identifiers from user-facing content, and adding explicit guidance on communication style.

Key Changes

• Standardized agent naming: Converted all agent references from kebab-case internal identifiers (e.g., market-data, osint-analyst) to proper Title Case display names (e.g., Market Data, OSINT Analyst) throughout the prompt and agent descriptions
• Removed layer labels: Eliminated L1/L2/L3/L4 layer designations from agent categories to simplify the mental model
• Added "Act first, talk second" principle: New leading rule emphasizing immediate delegation and action over description
• Enhanced communication guidelines:
  • Clarified that internal function/tool names should never be exposed to users
  • Added explicit instruction to use agent display names in natural language citations
  • New rules against showing internal identifiers and avoiding filler language
  • Emphasized leading with answers rather than process descriptions
• Improved rule clarity: Renumbered and reorganized rules for better readability, with specific examples of proper citation format
• Updated agent descriptions: Modified Market Analyst, OSINT Analyst, and Signals Analyst descriptions to reference display names instead of internal identifiers

Notable Details

The changes maintain all functional behavior while improving the user-facing experience by ensuring the assistant never exposes implementation details like store_get_notes() or osint-data to end users. The new "Act first, talk second" principle sets clear expectations for responsive, action-oriented assistance.

https://claude.ai/code/session_0111Nmtp4TyE8FMbmsm9PhHJ