cleveragents/cleveragents-core

Fork 3

UAT: `agents plan correct --mode revert` never restores actor reasoning state — `decisions` parameter not passed to `execute_correction`, `actor_state_ref` always empty #3108

New issue

Open

opened 2026-04-05 06:18:40 +00:00 by freemo · 1 comment

freemo commented

2026-04-05 06:18:40 +00:00

Owner

Summary

The agents plan correct --mode revert CLI command never passes the decisions mapping to execute_correction(). As a result, _extract_actor_state_ref() always returns "" and the actor_state_ref field in CorrectionResult is always empty. The reasoning rollback step (restoring the LangGraph actor's conversation history and graph state) is effectively disabled for all CLI-initiated corrections.

Metadata

Branch: fix/plan-correct-pass-decisions-to-execute-correction
Commit Message: fix(cli): pass decisions mapping to execute_correction for actor state restoration
Milestone: v3.2.0
Parent Epic: #394

Subtasks

Fetch the decisions mapping from DecisionService in correct_decision() in plan.py
Pass decisions to svc.execute_correction() call
Add Behave unit test verifying actor_state_ref is populated when decisions have context snapshots
Add Robot Framework integration test verifying reasoning rollback metadata is returned

Definition of Done

CorrectionResult.actor_state_ref is populated with the target decision's context_snapshot.actor_state_ref when the decision has one
All existing tests pass
All nox stages pass
Coverage e= 97%

Bug Details

What was tested

Code analysis of src/cleveragents/cli/commands/plan.py lines 3108 3113 and src/cleveragents/application/services/correction_service.py lines 889 914

Expected behavior (from spec § Correction Flow, Revert Mode)

Reasoning rollback: Extracts the actor_state_ref from the target decision's context_snapshot and returns it in the result so downstream consumers can restore the LangGraph actor's reasoning state.

The Decision should contain the actor_state_ref

Actual behavior

The CLI calls execute_correction without the decisions parameter:

# src/cleveragents/cli/commands/plan.py lines 3108-3113
result = svc.execute_correction(
    request.correction_id,
    decision_tree=decision_tree,
    influence_edges=influence_edges,
    # decisions= is MISSING — never passed
)

The _extract_actor_state_ref method returns "" when decisions is None:

# correction_service.py lines 907-914
@staticmethod
def _extract_actor_state_ref(target_decision_id, decisions):
    if decisions is None:
        return ""  # ← always returns "" because decisions is never passed
    ...

Code locations

src/cleveragents/cli/commands/plan.py, function correct_decision(), lines 3108 3113
src/cleveragents/application/services/correction_service.py, method _extract_actor_state_ref(), lines 889 914

Steps to reproduce

Create a plan with decisions that have context_snapshot.actor_state_ref set
Run: agents plan correct cdecision_ide --mode revert --guidance "test" --yes
Observe that actor_state_ref in the result is always "" regardless of the decision's context snapshot

Automated by CleverAgents Bot
Supervisor: UAT Testing | Agent: ca-new-issue-creator

## Summary The `agents plan correct --mode revert` CLI command never passes the `decisions` mapping to `execute_correction()`. As a result, `_extract_actor_state_ref()` always returns `""` and the `actor_state_ref` field in `CorrectionResult` is always empty. The reasoning rollback step (restoring the LangGraph actor's conversation history and graph state) is effectively disabled for all CLI-initiated corrections. ## Metadata - **Branch**: `fix/plan-correct-pass-decisions-to-execute-correction` - **Commit Message**: `fix(cli): pass decisions mapping to execute_correction for actor state restoration` - **Milestone**: v3.2.0 - **Parent Epic**: #394 ## Subtasks - [ ] Fetch the `decisions` mapping from `DecisionService` in `correct_decision()` in `plan.py` - [ ] Pass `decisions` to `svc.execute_correction()` call - [ ] Add Behave unit test verifying `actor_state_ref` is populated when decisions have context snapshots - [ ] Add Robot Framework integration test verifying reasoning rollback metadata is returned ## Definition of Done - `CorrectionResult.actor_state_ref` is populated with the target decision's `context_snapshot.actor_state_ref` when the decision has one - All existing tests pass - All nox stages pass - Coverage e= 97% ## Bug Details ### What was tested Code analysis of `src/cleveragents/cli/commands/plan.py` lines 3108 3113 and `src/cleveragents/application/services/correction_service.py` lines 889 914 ### Expected behavior (from spec § Correction Flow, Revert Mode) **Reasoning rollback**: Extracts the `actor_state_ref` from the target decision's `context_snapshot` and returns it in the result so downstream consumers can restore the LangGraph actor's reasoning state. The `Decision` should contain the actor_state_ref ### Actual behavior The CLI calls `execute_correction` without the `decisions` parameter: ```python # src/cleveragents/cli/commands/plan.py lines 3108-3113 result = svc.execute_correction( request.correction_id, decision_tree=decision_tree, influence_edges=influence_edges, # decisions= is MISSING — never passed ) ``` The `_extract_actor_state_ref` method returns `""` when `decisions` is None: ```python # correction_service.py lines 907-914 @staticmethod def _extract_actor_state_ref(target_decision_id, decisions): if decisions is None: return "" # ← always returns "" because decisions is never passed ... ``` ### Code locations - `src/cleveragents/cli/commands/plan.py`, function `correct_decision()`, lines 3108 3113 - `src/cleveragents/application/services/correction_service.py`, method `_extract_actor_state_ref()`, lines 889 914 ### Steps to reproduce 1. Create a plan with decisions that have `context_snapshot.actor_state_ref` set 2. Run: `agents plan correct cdecision_ide --mode revert --guidance "test" --yes` 3. Observe that `actor_state_ref` in the result is always `""` regardless of the decision's context snapshot --- **Automated by CleverAgents Bot** Supervisor: UAT Testing | Agent: ca-new-issue-creator

freemo added this to the v3.2.0 milestone

2026-04-05 06:18:47 +00:00

freemo added a new dependency

2026-04-05 06:20:10 +00:00

#394 Epic: Decision Framework

freemo added the

MoSCoW

Must have

label

2026-04-05 06:32:53 +00:00

freemo referenced this issue

2026-04-05 06:33:14 +00:00

UAT: execute_revert never supersedes decisions in the tree — CorrectionService not wired to DecisionService, affected decisions remain active after correction #3113

freemo commented

2026-04-05 06:33:32 +00:00

Author

Owner

Issue triaged by project owner:

State: Verified
Priority: Critical (keeping existing — reasoning rollback is a core correction feature)
Milestone: v3.2.0 (already assigned, keeping)
MoSCoW: Must Have — the spec requires reasoning rollback to extract actor_state_ref from the target decision's context snapshot. Without this, the LangGraph actor's state is never restored during corrections.
Parent Epic: #394 (Decision Framework)

Part of the correction-flow bug cluster (#3107, #3108, #3109, #3113, #3114).

Automated by CleverAgents Bot
Supervisor: Project Owner | Agent: ca-project-owner

Issue triaged by project owner: - **State**: Verified - **Priority**: Critical (keeping existing — reasoning rollback is a core correction feature) - **Milestone**: v3.2.0 (already assigned, keeping) - **MoSCoW**: Must Have — the spec requires reasoning rollback to extract `actor_state_ref` from the target decision's context snapshot. Without this, the LangGraph actor's state is never restored during corrections. - **Parent Epic**: #394 (Decision Framework) Part of the correction-flow bug cluster (#3107, #3108, #3109, #3113, #3114). --- **Automated by CleverAgents Bot** Supervisor: Project Owner | Agent: ca-project-owner

freemo referenced this issue

2026-04-05 06:33:33 +00:00

UAT: agents plan correct --mode=append is a stub — generates random ULIDs but does not create a real child plan or decision in the system #3114

freemo referenced this issue

2026-04-05 06:33:49 +00:00

UAT: agents plan correct JSON output format does not match spec — flat structure instead of nested correction/affected_subtree/sandbox_rollback/recompute/history #3107

freemo referenced this issue

2026-04-05 06:34:14 +00:00

[CA-AUTO] UAT Pool Supervisor — v3.2.0–v3.7.0 — Session Tracker #3094

freemo referenced this issue

2026-04-05 06:43:52 +00:00

[CA-AUTO] UAT Pool Supervisor — v3.2.0–v3.7.0 — Session Tracker #3094