Architecture of tauricresearch/tradingagents
tauricresearch/tradingagents is 195 source files holding 2,118 symbols across 4 modules, written mainly in Python, Dockerfile. The largest module is tests, with 90 files, and 4 files are entry points into the dependency graph.
How tradingagents fits together
Input enters as a ticker and a date, either through the interactive terminal tool in cli/main.py or by calling TradingAgentsGraph.propagate from tradingagents/graph/trading_graph.py. The tradingagents/graph package builds the LangGraph workflow that runs the analyst agents, then the research and risk debates, up to a portfolio decision. The tradingagents/agents directory writes the analyst reports a run starts from and holds the shared state and rating logic that later steps read. The tradingagents/dataflows package provides the helpers that market-data vendor calls rely on, such as settings, HTTP access and cache writes. Settings come from tradingagents/default_config.py, and the root main.py copies them before building the trading graph. A finished run is saved as a report tree, which includes complete_report.html.
Largest modules
- tests: 90 files and 11,300 lines of code
- tradingagents: 83 files and 8,253 lines of code
- cli: 10 files and 2,267 lines of code
- root: 2 files and 27 lines of code
Languages
By file count: Python 184 files, Dockerfile 1 file.
Documentation
Repowise documents tradingagents in 140 pages generated from the source, 14 of them written explanations of the repository and its modules. Module guides include:
- tradingagents/graph: Agent Graph Orchestration
- tradingagents/agents: Analyst Agents and Context
- root: Application Launcher
- tradingagents: Backtesting, Portfolio, and Reporting
- tradingagents/dataflows: Dataflow Config and Networking
More on tradingagents: summary and FAQ, codebase overview, code health.
Related guide: Best dependency graph tools for monorepos