fix: add missing ExecutorContext methods to AgentExecutor for human_input support. #6348
Open
VIVEK-311810 wants to merge 1 commit into
Open
fix: add missing ExecutorContext methods to AgentExecutor for human_input support. #6348VIVEK-311810 wants to merge 1 commit into
VIVEK-311810 wants to merge 1 commit into
Conversation
…nput support Add _invoke_loop, _ainvoke_loop, and _format_feedback_message to AgentExecutor so that the ExecutorContext / AsyncExecutorContext protocol is fully implemented, enabling human_input=True feedback cycles to work without AttributeError. Claude-Session: https://claude.ai/code/session_01PNsL23nAUz3bUy99Usc2Lz Co-authored-by: Claude <noreply@anthropic.com>
![corridor-security[bot]](https://avatars.githubusercontent.com/in/1177339?s=60&v=4){kind=small}
There was a problem hiding this comment.
Summary: This PR adds missing feedback-loop methods to AgentExecutor for synchronous and asynchronous human input handling. No exploitable security vulnerabilities were identified in the added code.
Risk: Low risk. The changes do not introduce new public endpoints, authentication or authorization logic, filesystem access, subprocess execution, SQL, or external network calls.
📝 WalkthroughWalkthroughAgentExecutor now formats human feedback messages and reruns the execution loop in synchronous and asynchronous paths. The rerun helpers clear completion state, invoke the executor again, and require the rerun to produce an ChangesHuman-feedback executor loop
Suggested reviewers
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@lib/crewai/src/crewai/experimental/agent_executor.py`:
- Around line 3191-3209: Reset the iteration counter before rerunning the
feedback loop, since _invoke_loop and _ainvoke_loop currently only clear
finalize/finished state and then call kickoff/kickoff_async, which leaves
self.state.iterations carrying over from the previous attempt. Update both
methods to mirror the per-run reset used by invoke() and invoke_async() by
resetting self.state.iterations before the rerun so check_max_iterations() and
force_final_answer don’t trigger immediately.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro Plus
Run ID: 068598bb-4e36-4306-a0e9-49ea1f31df43
📒 Files selected for processing (1)
lib/crewai/src/crewai/experimental/agent_executor.py
Comment on lines
+3191
to
+3209
| def _invoke_loop(self) -> AgentFinish: | ||
| """Re-run the agent flow loop for a feedback iteration (sync).""" | ||
| self._finalize_called = False | ||
| self.state.is_finished = False | ||
| self.kickoff() | ||
| result = self.state.current_answer | ||
| if not isinstance(result, AgentFinish): | ||
| raise RuntimeError("Agent execution ended without reaching a final answer.") | ||
| return result | ||
|
|
||
| async def _ainvoke_loop(self) -> AgentFinish: | ||
| """Re-run the agent flow loop for a feedback iteration (async).""" | ||
| self._finalize_called = False | ||
| self.state.is_finished = False | ||
| await self.kickoff_async() | ||
| result = self.state.current_answer | ||
| if not isinstance(result, AgentFinish): | ||
| raise RuntimeError("Agent execution ended without reaching a final answer.") | ||
| return result |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
Reset the iteration counter before rerunning after feedback.
kickoff()/kickoff_async() only clear Flow bookkeeping. They do not reset self.state.iterations, so a feedback rerun inherits the previous attempt's turn count and can immediately hit check_max_iterations()/force_final_answer without consuming the newly appended feedback. Mirror the per-run reset that invoke() and invoke_async() already do here.
Suggested fix
def _invoke_loop(self) -> AgentFinish:
"""Re-run the agent flow loop for a feedback iteration (sync)."""
self._finalize_called = False
self.state.is_finished = False
+ self.state.iterations = 0
+ self.state.current_answer = None
self.kickoff()
result = self.state.current_answer
if not isinstance(result, AgentFinish):
raise RuntimeError("Agent execution ended without reaching a final answer.")
return result
async def _ainvoke_loop(self) -> AgentFinish:
"""Re-run the agent flow loop for a feedback iteration (async)."""
self._finalize_called = False
self.state.is_finished = False
+ self.state.iterations = 0
+ self.state.current_answer = None
await self.kickoff_async()
result = self.state.current_answer
if not isinstance(result, AgentFinish):
raise RuntimeError("Agent execution ended without reaching a final answer.")
return result📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| def _invoke_loop(self) -> AgentFinish: | |
| """Re-run the agent flow loop for a feedback iteration (sync).""" | |
| self._finalize_called = False | |
| self.state.is_finished = False | |
| self.kickoff() | |
| result = self.state.current_answer | |
| if not isinstance(result, AgentFinish): | |
| raise RuntimeError("Agent execution ended without reaching a final answer.") | |
| return result | |
| async def _ainvoke_loop(self) -> AgentFinish: | |
| """Re-run the agent flow loop for a feedback iteration (async).""" | |
| self._finalize_called = False | |
| self.state.is_finished = False | |
| await self.kickoff_async() | |
| result = self.state.current_answer | |
| if not isinstance(result, AgentFinish): | |
| raise RuntimeError("Agent execution ended without reaching a final answer.") | |
| return result | |
| def _invoke_loop(self) -> AgentFinish: | |
| """Re-run the agent flow loop for a feedback iteration (sync).""" | |
| self._finalize_called = False | |
| self.state.is_finished = False | |
| self.state.iterations = 0 | |
| self.state.current_answer = None | |
| self.kickoff() | |
| result = self.state.current_answer | |
| if not isinstance(result, AgentFinish): | |
| raise RuntimeError("Agent execution ended without reaching a final answer.") | |
| return result | |
| async def _ainvoke_loop(self) -> AgentFinish: | |
| """Re-run the agent flow loop for a feedback iteration (async).""" | |
| self._finalize_called = False | |
| self.state.is_finished = False | |
| self.state.iterations = 0 | |
| self.state.current_answer = None | |
| await self.kickoff_async() | |
| result = self.state.current_answer | |
| if not isinstance(result, AgentFinish): | |
| raise RuntimeError("Agent execution ended without reaching a final answer.") | |
| return result |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@lib/crewai/src/crewai/experimental/agent_executor.py` around lines 3191 -
3209, Reset the iteration counter before rerunning the feedback loop, since
_invoke_loop and _ainvoke_loop currently only clear finalize/finished state and
then call kickoff/kickoff_async, which leaves self.state.iterations carrying
over from the previous attempt. Update both methods to mirror the per-run reset
used by invoke() and invoke_async() by resetting self.state.iterations before
the rerun so check_max_iterations() and force_final_answer don’t trigger
immediately.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds three methods to
AgentExecutorthat are required by theExecutorContextprotocolbut were never implemented when
AgentExecutorwas promoted to the default executor in 1.15.0.This caused
Task(human_input=True)to crash with anAttributeErrorfor all users who didnot explicitly set
executor_class.Problem
In 1.15.0
Agent.executor_classwas silently changed fromCrewAgentExecutortoAgentExecutoras the default (
agent/core.pyline 342). HoweverAgentExecutordoes not fully implement theExecutorContextprotocol thatSyncHumanInputProvider.handle_feedback()calls back on theexecutor after receiving user input.
Three methods required by the protocol were missing:
CrewAgentExecutorAgentExecutorbefore this fix_format_feedback_messageExecutorContextline 46_invoke_loopExecutorContextline 30_ainvoke_loopAsyncExecutorContextline 54The intent to support
human_input=Truewas clearly there —ask_for_human_inputstate,property,
_handle_human_feedback, and_ahandle_human_feedbackwere all implemented.These three callback methods were simply missed during the transition.
Error before this fix
Processing your feedback...
An unknown error occurred. Please check the details below.
Error details: 'AgentExecutor' object has no attribute '_format_feedback_message'
Changes
src/crewai/experimental/agent_executor.py: added_invoke_loop,_ainvoke_loop,_format_feedback_message— all using existing imports (format_message_for_llm,I18N_DEFAULT) already present in the fileTesting
Task(human_input=True)with defaultAgentExecutorno longer crashesCrewAgentExecutorin 1.14.3Linked Issue
Closes #6347
Summary by CodeRabbit
New Features
Bug Fixes