How to handle agent errors in LangChain
Quick answer
In
LangChain, handle agent errors by wrapping agent calls in try-except blocks to catch exceptions like APIError or ValueError. Implement retries with exponential backoff and fallback logic to ensure your app remains resilient when agents fail. ERROR TYPE
api_error ⚡ QUICK FIX
Add exponential backoff retry logic around your API call to handle
RateLimitError automatically.Why this happens
Agent errors in LangChain typically occur due to API rate limits, invalid inputs, or unexpected model responses. For example, calling agent.run() without handling exceptions can raise errors like openai.error.RateLimitError or ValueError if the input is malformed.
Example broken code:
from langchain.agents import initialize_agent, Tool
from langchain_openai import ChatOpenAI
import os
llm = OpenAI(model_name="gpt-4o", temperature=0, api_key=os.environ["OPENAI_API_KEY"])
tools = [Tool(name="Search", func=lambda x: "search result", description="Search tool")]
agent = initialize_agent(tools, llm, agent="zero-shot-react-description", verbose=True)
# This will raise an error if the input is invalid or API limit is hit
result = agent.run("Invalid input that causes error")
print(result) output
Traceback (most recent call last): ... openai.error.RateLimitError: You have exceeded your current quota, please check your plan and billing details.
The fix
Wrap the agent call in a try-except block and implement retries with exponential backoff to handle transient errors like rate limits. This prevents your app from crashing and allows graceful recovery.
import time
from langchain.agents import initialize_agent, Tool
from langchain_openai import ChatOpenAI
import os
llm = OpenAI(model_name="gpt-4o", temperature=0, api_key=os.environ["OPENAI_API_KEY"])
tools = [Tool(name="Search", func=lambda x: "search result", description="Search tool")]
agent = initialize_agent(tools, llm, agent="zero-shot-react-description", verbose=True)
max_retries = 3
for attempt in range(max_retries):
try:
result = agent.run("Your query here")
print(result)
break
except Exception as e:
print(f"Attempt {attempt + 1} failed: {e}")
if attempt < max_retries - 1:
time.sleep(2 ** attempt) # exponential backoff
else:
print("All retries failed. Please check your input or API limits.") output
Your query result here # or Attempt 1 failed: RateLimitError: ... Attempt 2 failed: RateLimitError: ... Attempt 3 failed: RateLimitError: ... All retries failed. Please check your input or API limits.
Preventing it in production
- Validate inputs before passing to the agent to avoid
ValueErroror malformed requests. - Use retry libraries like
tenacityfor robust retry policies with jitter and max delay. - Implement fallback logic, such as default responses or alternative tools, if the agent repeatedly fails.
- Monitor API usage and error rates to proactively adjust request rates or upgrade plans.
Key Takeaways
- Always wrap
agent.run()calls in try-except blocks to catch errors. - Use exponential backoff retries to handle transient API errors like rate limits.
- Validate inputs before sending to agents to prevent avoidable errors.
- Implement fallback strategies to maintain app resilience during agent failures.