langchain-ai/langgraph
1
0
Fork 0
mirror of https://github.com/langchain-ai/langgraph.git synced 2026-07-15 12:35:43 -04:00
Code Issues 314 Packages Projects Releases 17 Wiki Activity

[PR #4900] Fix: Resolve duplicate tool results in human approval flow #4397 #4128

New Issue
Closed
opened 2026-02-20 17:49:42 -05:00 by yindo · 0 comments
yindo commented 2026-02-20 17:49:42 -05:00
Owner
Copy Link
Copy Source

Original Pull Request: https://github.com/langchain-ai/langgraph/pull/4900

State: closed
Merged: No

Description

This PR fixes a critical bug where tools requiring human approval were generating multiple ToolMessage objects for single tool calls, causing "Too many tool_result blocks found" errors in the human approval workflow.

Problem

The existing implementation had several issues:

  • Flawed tool_result_handler logic with premature break statements
  • Graph flow problems causing repeated tool executions
  • Manual message manipulation that violated LangGraph design patterns
  • Missing duplicate prevention mechanisms

Solution

Root Cause Analysis

  • Tool Result Handler: The handler was not properly checking existing tool call IDs before processing
  • Graph Architecture: Manual message manipulation was bypassing LangGraph's built-in state management
  • Execution Flow: Missing safeguards allowed duplicate tool executions

Changes Made

  1. Fixed tool result handler logic - Properly validate existing tool call IDs before processing
  2. Added duplicate prevention - Implement checks in tool execution nodes to prevent multiple executions
  3. Simplified graph architecture - Removed problematic manual result handler
  4. Leveraged built-in functionality - Use add_messages for automatic deduplication following LangGraph patterns
  5. Comprehensive test coverage - Added test suite to reproduce and verify the fix

Testing

Test Files Added

  • libs/langgraph/tests/test_human_approval_tool_duplication_fix.py - Comprehensive test reproducing the bug and verifying fix
  • libs/langgraph/tests/test_human_approval_simple_fix.py - Simplified test case for basic validation
  • examples/human_in_the_loop/fixed_human_approval_example.py - Working example demonstrating the fix

Test Coverage

  • Reproduces original "Too many tool_result blocks found" error
  • Verifies single tool execution per approval
  • Tests backwards compatibility
  • Validates proper state management

Documentation

Documentation Added

  • docs/human_approval_tool_duplication_solution.md - Detailed technical solution documentation
  • SOLUTION_SUMMARY.md - High-level summary of changes and rationale

Backwards Compatibility

✅ This change maintains full backwards compatibility. Existing code using the human approval workflow will continue to work without modifications while benefiting from the duplicate prevention fixes.

Checklist

  • Follows LangGraph design principles for state management
  • Maintains backwards compatibility
  • Includes comprehensive test coverage
  • Adds appropriate documentation
  • Fixes critical bug without breaking changes
  • Uses built-in LangGraph patterns (add_messages) instead of manual manipulation

Related Issues

Fixes #4397

Additional Notes

The solution follows LangGraph's core philosophy of leveraging built-in state management rather than manual message manipulation. This approach not only fixes the immediate duplication bug but also makes the code more maintainable and aligned with framework best practices.

**Original Pull Request:** https://github.com/langchain-ai/langgraph/pull/4900 **State:** closed **Merged:** No --- ## Description This PR fixes a critical bug where tools requiring human approval were generating multiple `ToolMessage` objects for single tool calls, causing "Too many tool_result blocks found" errors in the human approval workflow. ## Problem The existing implementation had several issues: - Flawed `tool_result_handler` logic with premature break statements - Graph flow problems causing repeated tool executions - Manual message manipulation that violated LangGraph design patterns - Missing duplicate prevention mechanisms ## Solution ### Root Cause Analysis - **Tool Result Handler**: The handler was not properly checking existing tool call IDs before processing - **Graph Architecture**: Manual message manipulation was bypassing LangGraph's built-in state management - **Execution Flow**: Missing safeguards allowed duplicate tool executions ### Changes Made 1. **Fixed tool result handler logic** - Properly validate existing tool call IDs before processing 2. **Added duplicate prevention** - Implement checks in tool execution nodes to prevent multiple executions 3. **Simplified graph architecture** - Removed problematic manual result handler 4. **Leveraged built-in functionality** - Use `add_messages` for automatic deduplication following LangGraph patterns 5. **Comprehensive test coverage** - Added test suite to reproduce and verify the fix ## Testing ### Test Files Added - `libs/langgraph/tests/test_human_approval_tool_duplication_fix.py` - Comprehensive test reproducing the bug and verifying fix - `libs/langgraph/tests/test_human_approval_simple_fix.py` - Simplified test case for basic validation - `examples/human_in_the_loop/fixed_human_approval_example.py` - Working example demonstrating the fix ### Test Coverage - Reproduces original "Too many tool_result blocks found" error - Verifies single tool execution per approval - Tests backwards compatibility - Validates proper state management ## Documentation ### Documentation Added - `docs/human_approval_tool_duplication_solution.md` - Detailed technical solution documentation - `SOLUTION_SUMMARY.md` - High-level summary of changes and rationale ## Backwards Compatibility ✅ This change maintains full backwards compatibility. Existing code using the human approval workflow will continue to work without modifications while benefiting from the duplicate prevention fixes. ## Checklist - [x] Follows LangGraph design principles for state management - [x] Maintains backwards compatibility - [x] Includes comprehensive test coverage - [x] Adds appropriate documentation - [x] Fixes critical bug without breaking changes - [x] Uses built-in LangGraph patterns (`add_messages`) instead of manual manipulation ## Related Issues Fixes #4397 ## Additional Notes The solution follows LangGraph's core philosophy of leveraging built-in state management rather than manual message manipulation. This approach not only fixes the immediate duplication bug but also makes the code more maintainable and aligned with framework best practices.
yindo added the pull-request label 2026-02-20 17:49:42 -05:00
yindo closed this issue 2026-02-20 17:49:42 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
agents
awaiting author revision
bug
change
chat models
checkpointer
ci
code quality
code quality
dependencies
documentation
drawing
duplicate
enhancement
good first issue
help wanted
invalid
maintainer
meta
missing information
not maintained by langchain
open-swe-auto
open-swe-max
open-swe-max-auto
pending
performance
pull-request
Mirrored from GitHub Pull Request
 question
requires design
sdk
server
stale
studio
tests
type checking
usage question
v0.5
v1.1
No Label pull-request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langchain-ai/langgraph#4128
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?