TradingAgents/tradingagents
dtarkent2-sys 806263defd fix: process all chunk messages + harden memory score normalization
Backport from upstream TauricResearch/TradingAgents fa4d01c2:
  - cli/main.py: iterate all messages in each chunk (not just the last);
    dedupe across chunks with a processed-IDs set instead of tracking
    only the most recent message. Prevents dropped tool-call logs when
    the graph streams multiple messages per chunk.
  - memory.py: guard max() against empty scores and return a float to
    avoid ValueError on first-call paths before documents are added.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-18 15:09:34 -04:00
..
agents fix: process all chunk messages + harden memory score normalization 2026-04-18 15:09:34 -04:00
dataflows WIP: local TradingAgents customizations through 2026-04-13 2026-04-13 22:01:00 -04:00
graph WIP: local TradingAgents customizations through 2026-04-13 2026-04-13 22:01:00 -04:00
llm_clients Round 1: import upstream additions (TauricResearch/TradingAgents) 2026-04-13 22:20:16 -04:00
__init__.py Round 1: import upstream additions (TauricResearch/TradingAgents) 2026-04-13 22:20:16 -04:00
default_config.py chore: default deep_think_llm + quick_think_llm to glm-5.1:cloud 2026-04-13 10:06:37 -04:00
models.py fix: aggressive JSON fallback for LLMs that don't support structured output 2026-04-02 15:39:08 -04:00