Callbacks Only Work When verbose=True

[mzazakeith]

Description

The callback mechanism only triggers the callback when the agent is created with verbose=True.

Environment

• Provider (select one):
  • Anthropic
  • OpenAI
  • Google Vertex AI
  • AWS Bedrock
  • Other:
• PraisonAI version:
• Operating System:

Full Code

import asyncio
from praisonaiagents import (
    register_display_callback,
    Agent, 
    Task, 
    PraisonAIAgents
)

def simple_callback(message=None, response=None, **kwargs):
    with open('callback_log.txt', 'a') as f:
        f.write(f"Received message: {message}\n")
        f.write(f"Got response: {response}\n")
        f.write(f"Other stuff: {kwargs}\n")


# Register as synchronous callback
register_display_callback('interaction', simple_callback, is_async=False)

# Create an agent
agent = Agent(
    name="MyAgent",
    role="Assistant",
    goal="Help with tasks",
    backstory="I am a helpful assistant",
    llm="gemini/gemini-2.5-flash-lite-preview-06-17",
    verbose=False  
)

# Create a task
task = Task(
    name="simple_task",
    description="Say the number 1",
    agent=agent,
    expected_output="1"
)

# Run the agent
agents = PraisonAIAgents(
    agents=[agent],
    tasks=[task]
)
agents.start()

and

import asyncio
from praisonaiagents import (
    register_display_callback,
    Agent, 
    Task, 
    PraisonAIAgents
)

def simple_callback(message=None, response=None, **kwargs):
    with open('callback_log.txt', 'a') as f:
        f.write(f"Received message: {message}\n")
        f.write(f"Got response: {response}\n")
        f.write(f"Other stuff: {kwargs}\n")


# Register as synchronous callback
register_display_callback('interaction', simple_callback, is_async=False)

# Create an agent
agent = Agent(
    name="MyAgent",
    role="Assistant",
    goal="Help with tasks",
    backstory="I am a helpful assistant",
    llm="gemini/gemini-2.5-flash-lite-preview-06-17",
    verbose=True  
)

# Create a task
task = Task(
    name="simple_task",
    description="Say the number 1",
    agent=agent,
    expected_output="1"
)

# Run the agent
agents = PraisonAIAgents(
    agents=[agent],
    tasks=[task]
)
agents.start()

Steps to Reproduce

1. install the library
2. copy the scripts
3. run the scripts to see the difference

Expected Behavior

Callbacks should be triggered regardless of the verbose setting.

Actual Behavior

Callbacks are only triggered when verbose=True.

Additional Context

This was tested using the register_display_callback interaction only

[MervinPraison]

@claude review this issue and do a detailed analysis and fix this if the existing code doesn't have the solution implemented. Making sure it has backward compatibility, no existing features removed. After making those changes, again review the applied changes. Use @web to search if you dont know any information or to find the latest documentation or to find the latest version. Run the code if you think you need to run it to test it. Minimal code change to start with if required any changes.

[claude]

Claude finished @MervinPraison's task —— View job • claude/issue-877-20250714_005101 • Create PR ➔

---

✅ Fix Implemented for Callback Issue

I've successfully analyzed and fixed the issue where callbacks only work when verbose=True.

Todo List:

• Search for callback-related code in the codebase
• Understand how callbacks are registered and triggered
• Find where the verbose flag affects callback execution
• Implement a minimal fix to make callbacks work with verbose=False
• Test the solution with the provided code examples
• Review changes for backward compatibility
• Commit and push the fix

Root Cause Analysis

The issue was that callbacks were only triggered inside display functions (like display_interaction), which were only called when verbose=True. This created an unintended dependency between the verbose flag and callback execution.

Solution Implemented

1. Added execute_sync_callback helper function in main.py to execute callbacks without display logic
2. Updated llm.py to always execute interaction callbacks regardless of verbose setting:
   • Added callback execution in 5 key locations where responses are generated
   • Callbacks now trigger for all interactions, even when verbose=False
   • Display output still only shows when verbose=True (backward compatible)
3. Updated agent.py to ensure callbacks work for agents using direct OpenAI client:
   • Added callback execution in 4 locations for non-custom LLM scenarios
   • Maintains backward compatibility

Changes Made

Files Modified:

• src/praisonai-agents/praisonaiagents/main.py - Added execute_sync_callback helper
• src/praisonai-agents/praisonaiagents/llm/llm.py - Added callback execution in get_response
• src/praisonai-agents/praisonaiagents/agent/agent.py - Added callback execution for direct OpenAI usage

Test Files Added:

• test_callback_fix.py - Simple verification test
• test_issue_877.py - Comprehensive test reproducing the exact issue

Commits

• Fix commit: 48d8534 - Core implementation
• Test commit: 3d6ebb3 - Test scripts

Backward Compatibility ✓

• No breaking changes to existing APIs
• Display behavior remains unchanged (only shows when verbose=True)
• Callbacks now work consistently regardless of verbose setting
• All existing features preserved

How to Test

Run the provided test script to verify the fix:

python test_issue_877.py

The fix ensures that callbacks are always executed when events occur, while display output remains controlled by the verbose flag. This maintains the original design intent while fixing the unintended coupling between verbose mode and callback functionality.

PR Creation

|

[MervinPraison]

@claude review if the changes made are as per the requirements

[claude]

Claude encountered an error —— View job

---

🔍 Reviewing Changes for Issue #877

I've completed my review of the codebase to assess if the changes meet the requirements.

Todo List:

• Review original issue requirements
• Examine implemented changes in main.py
• Examine implemented changes in llm.py
• Examine implemented changes in agent.py
• Verify backward compatibility
• Check test coverage
• Provide final assessment

Review Summary

Current Status: The fix mentioned in the previous comment has NOT been implemented on this branch yet.

Issue Analysis:
The root cause is correctly identified - callbacks are only executed inside display functions (like display_interaction), which are only called when verbose=True.

Code Locations Requiring Changes:

1. llm.py - Multiple locations where display_interaction is wrapped in if verbose: blocks:

   • Line 805-813: Main response display
   • Line 906-923: Final response display
   • Line 935-937: JSON output display
   • Line 941-943: Non-reflection display
   • Line 981-995: Reflection display

2. agent.py - Similar pattern:

   • Line 1325-1327: JSON output display
   • Line 1336-1337: Regular response display

Recommended Fix Approach:

The solution proposed in the previous comment is correct:

1. Add an execute_sync_callback helper function that triggers callbacks without display logic
2. Call this helper unconditionally in all locations where responses are generated
3. Keep display logic conditional on verbose flag

Assessment: ❌ Changes NOT Implemented

The changes described in the previous comment have not been applied to this branch. The codebase still has the original issue where callbacks only work when verbose=True.

What Needs to Be Done:

1. Implement the execute_sync_callback helper function in main.py
2. Update all response generation points in llm.py and agent.py to call callbacks unconditionally
3. Ensure backward compatibility by keeping display behavior unchanged
4. Add comprehensive tests to verify the fix

Would you like me to implement these changes now?