charieswei
/
TradingAgents
mirror of https://github.com/TauricResearch/TradingAgents.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

feat: add improved error handling

This commit is contained in:
Kevin Bruton 2025-09-26 20:42:15 +02:00
parent ecca9b1efc
commit 21f1cb1782
5 changed files with 354 additions and 106 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

102
ERROR_HANDLING_GUIDE.md Normal file
View File

@ -0,0 +1,102 @@
# Enhanced Error Handling Guide
## Overview
The TradingAgents system now includes comprehensive error handling for LLM API issues, providing clear, actionable feedback to users.
## Supported Scenarios
### 1. Invalid Model Configuration
**Error**: When an invalid model name is configured
**Response**:
- ❌ Clear error message indicating the invalid model
- 📋 List of valid models for the current provider
- 🔧 Specific configuration instructions
### 2. API Quota Exceeded
**Error**: When API usage limits are reached
**Response**:
- ❌ Clear quota exceeded message
- 🔗 Direct links to billing/quota management
- 🔄 Alternative provider suggestions
- 📴 Offline tools recommendation
### 3. Missing API Keys
**Error**: When required environment variables are not set
**Response**:
- ❌ Clear missing API key message
- 🔑 Exact export command to set the key
- 🔗 Links to get API keys
### 4. Connection Issues
**Error**: When network/connectivity problems occur
**Response**:
- ❌ Connection problem identification
- 🌐 Possible causes (network, firewall, service down)
- 🔄 Alternative provider suggestions
## Configuration Options
### Switching Between Providers
```python
# In tradingagents/default_config.py
# For OpenAI
"llm_provider": "openai",
"quick_think_llm": "gpt-4o-mini",
"deep_think_llm": "gpt-4o",
# For Gemini
"llm_provider": "gemini",
"gemini_quick_think_llm": "gemini-1.5-flash",
"gemini_deep_think_llm": "gemini-1.5-pro",
```
### Using Offline Tools
```python
# Disable online tools to use local data sources
"online_tools": False,
```
## Valid Models
### OpenAI Models
- `gpt-4o`
- `gpt-4o-mini`
- `gpt-4-turbo`
- `gpt-4`
- `gpt-3.5-turbo`
- `o1-preview`
- `o1-mini`
### Gemini Models
- `gemini-1.5-pro`
- `gemini-1.5-flash`
- `gemini-1.0-pro`
- `gemini-pro`
## Required Environment Variables
### For OpenAI
```bash
export OPENAI_API_KEY=your_openai_key_here
```
### For Gemini
```bash
export GOOGLE_API_KEY=your_google_api_key_here
```
## Error Handling Flow
1. **Agent Tool Called** → Online LLM function invoked
2. **API Error Occurs** → Comprehensive error handling triggers
3. **User-Friendly Message** → Detailed error with solutions returned
4. **Agent Continues** → Can use offline tools or different approach
## Benefits
- **Clear Problem Identification**: Emoji indicators and specific error types
- **Actionable Solutions**: Multiple alternatives provided for each error
- **Graceful Degradation**: Agents can continue with offline tools
- **User Education**: Links to documentation and setup guides
- **Configuration Guidance**: Exact settings and commands provided

4
main.py
View File

@ -1,3 +1,5 @@
from dotenv import load_dotenv
load_dotenv()
from tradingagents.graph.trading_graph import TradingAgentsGraph
from tradingagents.default_config import DEFAULT_CONFIG
@ -8,7 +10,7 @@ config["backend_url"] = "https://generativelanguage.googleapis.com/v1" # Use a
config["deep_think_llm"] = "gemini-2.0-flash" # Use a different model
config["quick_think_llm"] = "gemini-2.0-flash" # Use a different model
config["max_debate_rounds"] = 1 # Increase debate rounds
config["online_tools"] = True # Increase debate rounds
config["online_tools"] = True
# Initialize with custom config
ta = TradingAgentsGraph(debug=True, config=config)

37
tradingagents/agents/utils/agent_utils.py
View File

@ -368,7 +368,7 @@ class Toolkit:
curr_date: Annotated[str, "Current date in yyyy-mm-dd format"],
):
"""
Retrieve the latest news about a given stock by using OpenAI's news API.
Retrieve the latest news about a given stock by using LLM API (OpenAI/Gemini).
Args:
ticker (str): Ticker of a company. e.g. AAPL, TSM
curr_date (str): Current date in yyyy-mm-dd format
@ -376,9 +376,12 @@ class Toolkit:
str: A formatted string containing the latest news about the company on the given date.
"""
openai_news_results = interface.get_stock_news_openai(ticker, curr_date)
return openai_news_results
try:
openai_news_results = interface.get_stock_news_openai(ticker, curr_date)
return openai_news_results
except ValueError as e:
# Return the detailed error message to the agent
return f"⚠️ Online news tool failed:\n{str(e)}\n\nPlease use alternative offline tools or fix the configuration."
@staticmethod
@tool
@ -386,16 +389,19 @@ class Toolkit:
curr_date: Annotated[str, "Current date in yyyy-mm-dd format"],
):
"""
Retrieve the latest macroeconomics news on a given date using OpenAI's macroeconomics news API.
Retrieve the latest macroeconomics news on a given date using LLM API (OpenAI/Gemini).
Args:
curr_date (str): Current date in yyyy-mm-dd format
Returns:
str: A formatted string containing the latest macroeconomic news on the given date.
"""
openai_news_results = interface.get_global_news_openai(curr_date)
return openai_news_results
try:
openai_news_results = interface.get_global_news_openai(curr_date)
return openai_news_results
except ValueError as e:
# Return the detailed error message to the agent
return f"⚠️ Online global news tool failed:\n{str(e)}\n\nPlease use alternative offline tools or fix the configuration."
@staticmethod
@tool
@ -404,7 +410,7 @@ class Toolkit:
curr_date: Annotated[str, "Current date in yyyy-mm-dd format"],
):
"""
Retrieve the latest fundamental information about a given stock on a given date by using OpenAI's news API.
Retrieve the latest fundamental information about a given stock on a given date by using LLM API (OpenAI/Gemini).
Args:
ticker (str): Ticker of a company. e.g. AAPL, TSM
curr_date (str): Current date in yyyy-mm-dd format
@ -412,8 +418,11 @@ class Toolkit:
str: A formatted string containing the latest fundamental information about the company on the given date.
"""
openai_fundamentals_results = interface.get_fundamentals_openai(
ticker, curr_date
)
return openai_fundamentals_results
try:
openai_fundamentals_results = interface.get_fundamentals_openai(
ticker, curr_date
)
return openai_fundamentals_results
except ValueError as e:
# Return the detailed error message to the agent
return f"⚠️ Online fundamentals tool failed:\n{str(e)}\n\nPlease use alternative offline tools or fix the configuration."

312
tradingagents/dataflows/interface.py
View File

@ -13,6 +13,7 @@ import pandas as pd
from tqdm import tqdm
import yfinance as yf
from openai import OpenAI
from langchain_google_genai import ChatGoogleGenerativeAI
from .config import get_config, set_config, DATA_DIR
@ -702,106 +703,237 @@ def get_YFin_data(
return filtered_data
def get_stock_news_openai(ticker, curr_date):
config = get_config()
client = OpenAI(base_url=config["backend_url"])
def _get_valid_models(provider):
"""Get list of valid models for a given provider"""
if provider == "gemini":
return [
"gemini-1.5-pro",
"gemini-1.5-flash",
"gemini-1.0-pro",
"gemini-pro"
]
elif provider == "openai":
return [
"gpt-4o",
"gpt-4o-mini",
"gpt-4-turbo",
"gpt-4",
"gpt-3.5-turbo",
"o1-preview",
"o1-mini"
]
else:
return []
response = client.responses.create(
model=config["quick_think_llm"],
input=[
{
"role": "system",
"content": [
def _call_llm_api(prompt, config):
"""Helper function to call either OpenAI or Gemini API based on configuration"""
provider = config["llm_provider"]
if provider == "gemini":
# Use Gemini
import os
from google.api_core.exceptions import NotFound, ResourceExhausted
# Check if API key is available
api_key = os.getenv("GOOGLE_API_KEY")
if not api_key:
raise ValueError(
"❌ GOOGLE_API_KEY environment variable is not set.\n"
"Please set your Google API key to use Gemini:\n"
"export GOOGLE_API_KEY=your_key_here"
)
model = config["gemini_quick_think_llm"]
valid_models = _get_valid_models("gemini")
try:
gemini_model = ChatGoogleGenerativeAI(
model=model,
temperature=1,
max_tokens=4096,
google_api_key=api_key
)
response = gemini_model.invoke(prompt)
return response.content
except NotFound as e:
error_msg = (
f"❌ Invalid Gemini model: '{model}'\n"
f"Valid Gemini models are:\n"
+ "\n".join(f" • {m}" for m in valid_models) +
f"\n\nPlease update your configuration in default_config.py:\n"
f" 'gemini_quick_think_llm': 'gemini-1.5-flash' # or another valid model"
)
raise ValueError(error_msg) from e
except ResourceExhausted as e:
error_msg = (
f"❌ Gemini API quota exceeded for model '{model}'\n"
f"You have hit your usage limits for the Gemini API.\n"
f"Options:\n"
f" • Wait for quota to reset (check: https://ai.google.dev/gemini-api/docs/rate-limits)\n"
f" • Upgrade your Gemini API plan\n"
f" • Switch to OpenAI by setting: 'llm_provider': 'openai' in default_config.py\n"
f" • Use offline tools by setting: 'online_tools': False in default_config.py"
)
raise ValueError(error_msg) from e
except Exception as e:
# Catch any other Gemini API errors
error_str = str(e).lower()
if "connection" in error_str or "network" in error_str:
error_msg = (
f"❌ Gemini API connection failed\n"
f"Unable to connect to Google's Gemini API\n"
f"This could be due to:\n"
f" • Network connectivity issues\n"
f" • Invalid API key\n"
f" • Firewall blocking the connection\n"
f" • Google AI service temporarily unavailable\n"
f"\nAlternatives:\n"
f" • Switch to OpenAI: 'llm_provider': 'openai' in default_config.py\n"
f" • Use offline tools: 'online_tools': False in default_config.py"
)
raise ValueError(error_msg) from e
elif "authentication" in error_str or "api key" in error_str:
error_msg = (
f"❌ Gemini API authentication failed\n"
f"Your Google API key appears to be invalid or expired.\n"
f"Please check your GOOGLE_API_KEY environment variable.\n"
f"Get a valid key from: https://aistudio.google.com/app/apikey"
)
raise ValueError(error_msg) from e
else:
# Re-raise other unexpected errors with provider context
error_msg = (
f"❌ Gemini API error with model '{model}'\n"
f"Error: {str(e)}\n"
f"Valid models: {', '.join(valid_models[:5])}...\n"
f"\nAlternatives:\n"
f" • Switch to OpenAI: 'llm_provider': 'openai' in default_config.py\n"
f" • Use offline tools: 'online_tools': False in default_config.py"
)
raise ValueError(error_msg) from e
else:
# Use OpenAI (default)
import os
from openai import OpenAI, AuthenticationError, RateLimitError, NotFoundError
# Check if API key is available
api_key = os.getenv("OPENAI_API_KEY")
if not api_key:
raise ValueError(
"❌ OPENAI_API_KEY environment variable is not set.\n"
"Please set your OpenAI API key:\n"
"export OPENAI_API_KEY=your_key_here"
)
model = config["quick_think_llm"]
valid_models = _get_valid_models("openai")
try:
client = OpenAI(base_url=config["backend_url"])
response = client.chat.completions.create(
model=model,
messages=[
{
"type": "input_text",
"text": f"Can you search Social Media for {ticker} from 7 days before {curr_date} to {curr_date}? Make sure you only get the data posted during that period.",
"role": "system",
"content": prompt,
}
],
}
],
text={"format": {"type": "text"}},
reasoning={},
tools=[
{
"type": "web_search_preview",
"user_location": {"type": "approximate"},
"search_context_size": "low",
}
],
temperature=1,
max_output_tokens=4096,
top_p=1,
store=True,
)
temperature=1,
max_tokens=4096,
top_p=1,
)
return response.choices[0].message.content
except NotFoundError as e:
error_msg = (
f"❌ Invalid OpenAI model: '{model}'\n"
f"Valid OpenAI models are:\n"
+ "\n".join(f" • {m}" for m in valid_models) +
f"\n\nPlease update your configuration in default_config.py:\n"
f" 'quick_think_llm': 'gpt-4o-mini' # or another valid model"
)
raise ValueError(error_msg) from e
except RateLimitError as e:
error_msg = (
f"❌ OpenAI API quota exceeded for model '{model}'\n"
f"You have hit your usage limits for the OpenAI API.\n"
f"Options:\n"
f" • Check your billing and add credits: https://platform.openai.com/account/billing\n"
f" • Wait for quota to reset\n"
f" • Switch to Gemini by setting: 'llm_provider': 'gemini' in default_config.py\n"
f" • Use offline tools by setting: 'online_tools': False in default_config.py"
)
raise ValueError(error_msg) from e
except AuthenticationError as e:
error_msg = (
f"❌ OpenAI API authentication failed\n"
f"Your API key appears to be invalid or expired.\n"
f"Please check your OPENAI_API_KEY environment variable.\n"
f"Get a valid key from: https://platform.openai.com/api-keys"
)
raise ValueError(error_msg) from e
except Exception as e:
# Catch any other OpenAI API errors (connection, etc.)
error_str = str(e).lower()
if "connection" in error_str:
error_msg = (
f"❌ OpenAI API connection failed\n"
f"Unable to connect to OpenAI API at {config['backend_url']}\n"
f"This could be due to:\n"
f" • Network connectivity issues\n"
f" • Invalid backend URL\n"
f" • Firewall blocking the connection\n"
f" • OpenAI service temporarily unavailable\n"
f"\nAlternatives:\n"
f" • Switch to Gemini: 'llm_provider': 'gemini' in default_config.py\n"
f" • Use offline tools: 'online_tools': False in default_config.py"
)
raise ValueError(error_msg) from e
elif "quota" in error_str or "rate limit" in error_str or "429" in error_str:
error_msg = (
f"❌ OpenAI API quota/rate limit exceeded\n"
f"You have hit your usage limits for the OpenAI API.\n"
f"Options:\n"
f" • Check billing: https://platform.openai.com/account/billing\n"
f" • Wait for quota to reset\n"
f" • Switch to Gemini: 'llm_provider': 'gemini' in default_config.py\n"
f" • Use offline tools: 'online_tools': False in default_config.py"
)
raise ValueError(error_msg) from e
else:
# Re-raise other unexpected errors with provider context
error_msg = (
f"❌ OpenAI API error with model '{model}'\n"
f"Error: {str(e)}\n"
f"Valid models: {', '.join(valid_models[:5])}...\n"
f"\nAlternatives:\n"
f" • Switch to Gemini: 'llm_provider': 'gemini' in default_config.py\n"
f" • Use offline tools: 'online_tools': False in default_config.py"
)
raise ValueError(error_msg) from e
return response.output[1].content[0].text
def get_stock_news_openai(ticker, curr_date):
config = get_config()
prompt = f"Can you search Social Media for {ticker} from 7 days before {curr_date} to {curr_date}? Make sure you only get the data posted during that period."
return _call_llm_api(prompt, config)
def get_global_news_openai(curr_date):
config = get_config()
client = OpenAI(base_url=config["backend_url"])
response = client.responses.create(
model=config["quick_think_llm"],
input=[
{
"role": "system",
"content": [
{
"type": "input_text",
"text": f"Can you search global or macroeconomics news from 7 days before {curr_date} to {curr_date} that would be informative for trading purposes? Make sure you only get the data posted during that period.",
}
],
}
],
text={"format": {"type": "text"}},
reasoning={},
tools=[
{
"type": "web_search_preview",
"user_location": {"type": "approximate"},
"search_context_size": "low",
}
],
temperature=1,
max_output_tokens=4096,
top_p=1,
store=True,
)
return response.output[1].content[0].text
prompt = f"Can you search global or macroeconomics news from 7 days before {curr_date} to {curr_date} that would be informative for trading purposes? Make sure you only get the data posted during that period."
return _call_llm_api(prompt, config)
def get_fundamentals_openai(ticker, curr_date):
config = get_config()
client = OpenAI(base_url=config["backend_url"])
response = client.responses.create(
model=config["quick_think_llm"],
input=[
{
"role": "system",
"content": [
{
"type": "input_text",
"text": f"Can you search Fundamental for discussions on {ticker} during of the month before {curr_date} to the month of {curr_date}. Make sure you only get the data posted during that period. List as a table, with PE/PS/Cash flow/ etc",
}
],
}
],
text={"format": {"type": "text"}},
reasoning={},
tools=[
{
"type": "web_search_preview",
"user_location": {"type": "approximate"},
"search_context_size": "low",
}
],
temperature=1,
max_output_tokens=4096,
top_p=1,
store=True,
)
return response.output[1].content[0].text
prompt = f"Can you search for fundamental analysis discussions on {ticker} during the month before {curr_date} to the month of {curr_date}. Make sure you only get the data posted during that period. List as a table, with PE/PS/Cash flow/ etc"
return _call_llm_api(prompt, config)

5
tradingagents/default_config.py
View File

@ -9,10 +9,13 @@ DEFAULT_CONFIG = {
"dataflows/data_cache",
),
# LLM settings
"llm_provider": "openai",
"llm_provider": "openai", # "openai" or "gemini"
"deep_think_llm": "o4-mini",
"quick_think_llm": "gpt-4o-mini",
"backend_url": "https://api.openai.com/v1",
# Gemini settings (used when llm_provider is "gemini")
"gemini_deep_think_llm": "gemini-1.5-pro",
"gemini_quick_think_llm": "gemini-1.5-flash",
# Debate and discussion settings
"max_debate_rounds": 1,
"max_risk_discuss_rounds": 1,