trader.chat package

The chat CLI’s Core Service Layer (requirements §4, §13).

A second entry point into the same Core, alongside the trading loop — chat must not need the daemon running, and the daemon must not need chat running (requirements §4). It reads the same SQLite tables the trading loop writes (trades, decisions, account_snapshots) and the same config/strategies.yaml, and talks to the local model through the existing LlmProvider Protocol — no new provider, and Anthropic stays deferred (see docs/ideas.md).

Deliberately no re-exports, matching every other package in this codebase (strategies, brokers, marketdata, news, persistence, llm have empty __init__.py files; discovery, execution and risk carry a docstring and nothing else) — import from the submodule you need: trader.chat.service.ChatService, trader.chat.intent, trader.chat.prompt.

Submodules

• trader.chat.backtest_format module
  • format_backtest_result()
• trader.chat.context module
  • decision_context()
  • performance_context()
  • positions_context()
  • trades_context()
• trader.chat.intent module
  • Intent
    • Intent.GENERAL
    • Intent.EXPLAIN_DECISION
    • Intent.BACKTEST
    • Intent.STRATEGY_REVIEW
  • ParsedIntent
    • ParsedIntent.intent
    • ParsedIntent.raw_message
    • ParsedIntent.symbol
    • ParsedIntent.decision_id
    • ParsedIntent.on_date
    • ParsedIntent.strategy_id
    • ParsedIntent.ticker
    • ParsedIntent.start
    • ParsedIntent.end
  • classify()
• trader.chat.prompt module
  • build_chat_messages()
  • build_explain_decision_messages()
  • build_strategy_review_messages()
  • parse_chat_response()
• trader.chat.service module
  • ChatService
    • ChatService.__init__()
    • ChatService.handle()