cleveragents/cleveragents-core

Fork 3

UAT: `agents plan correct --mode=append` spawns synthetic in-memory child plan but never persists guidance to decision tree or plan lifecycle #6149

New issue

Open

opened 2026-04-09 15:34:52 +00:00 by HAL9000 · 1 comment

HAL9000 commented

2026-04-09 15:34:52 +00:00

Owner

Summary

agents plan correct --mode=append calls execute_append() which generates a synthetic child plan ULID and new decision ULID entirely in-memory, but never persists them to the database, never creates a real child plan via SubplanService, and never records the guidance in the decision tree. The spec requires "Append correction recorded; no subtree recomputation triggered" — but nothing is actually recorded.

What Was Tested

Code-level analysis of CorrectionService.execute_append() against spec §Correction Model — Append Mode and v3.3.0 Acceptance Criterion #8.

Expected Behavior (from spec)

v3.3.0 Acceptance Criterion #8:

agents plan correct --mode=append adds guidance without recomputing | Append correction recorded; no subtree recomputation triggered

Spec §agents plan correct (append mode output):

Additional work: appended as new child plan

The spec requires that append mode:

Records the correction in correction_attempts table
Creates a real child plan (or at minimum a subplan_spawn decision) linked to the guidance
Does NOT trigger subtree recomputation

Actual Behavior

execute_append() generates synthetic IDs in-memory and returns without any persistence:

# src/cleveragents/application/services/correction_service.py
def execute_append(self, correction_id: str) -> CorrectionResult:
    ...
    try:
        child_plan_id = str(ULID())       # synthetic, never persisted
        new_decision_id = str(ULID())     # synthetic, never persisted
        
        result = CorrectionResult(
            correction_id=correction_id,
            status=CorrectionStatus.APPLIED,
            new_decisions=[new_decision_id],
            spawned_child_plan_id=child_plan_id,
        )
        ...

No calls to:

UnitOfWork.correction_attempts.create() — correction not persisted
SubplanService.spawn_subplan() — no real child plan created
DecisionService.create_decision() — no subplan_spawn decision recorded
Any plan lifecycle service to register the child plan

The CLI output shows new_decisions: [<ULID>] and spawned_child_plan_id: <ULID> but these are phantom IDs that don't exist in any database table.

Steps to Reproduce

Run agents plan correct --mode append -g "Add rate limiting" <DECISION_ID> --yes
Observe: command reports success with spawned_child_plan_id
Run agents plan status <spawned_child_plan_id>
Observe: "Plan not found" — the child plan was never created

Code Locations

src/cleveragents/application/services/correction_service.py — execute_append() lines ~490-530
src/cleveragents/application/services/subplan_service.py — SubplanService.spawn_subplan() exists but is never called from execute_append()

Spec References

§Correction Model — Append Mode
v3.3.0 Acceptance Criterion #8: "Append correction recorded; no subtree recomputation triggered"
v3.3.0 Acceptance Criterion #9: "Correction flow functional (plan correct --mode revert and --mode append)"

Automated by CleverAgents Bot
Supervisor: UAT Testing | Agent: uat-tester

## Summary `agents plan correct --mode=append` calls `execute_append()` which generates a synthetic child plan ULID and new decision ULID entirely in-memory, but never persists them to the database, never creates a real child plan via `SubplanService`, and never records the guidance in the decision tree. The spec requires "Append correction recorded; no subtree recomputation triggered" — but nothing is actually recorded. ## What Was Tested Code-level analysis of `CorrectionService.execute_append()` against spec §Correction Model — Append Mode and v3.3.0 Acceptance Criterion #8. ## Expected Behavior (from spec) **v3.3.0 Acceptance Criterion #8:** > `agents plan correct --mode=append` adds guidance without recomputing | Append correction recorded; no subtree recomputation triggered **Spec §agents plan correct (append mode output):** ``` Additional work: appended as new child plan ``` The spec requires that append mode: 1. Records the correction in `correction_attempts` table 2. Creates a real child plan (or at minimum a `subplan_spawn` decision) linked to the guidance 3. Does NOT trigger subtree recomputation ## Actual Behavior `execute_append()` generates synthetic IDs in-memory and returns without any persistence: ```python # src/cleveragents/application/services/correction_service.py def execute_append(self, correction_id: str) -> CorrectionResult: ... try: child_plan_id = str(ULID()) # synthetic, never persisted new_decision_id = str(ULID()) # synthetic, never persisted result = CorrectionResult( correction_id=correction_id, status=CorrectionStatus.APPLIED, new_decisions=[new_decision_id], spawned_child_plan_id=child_plan_id, ) ... ``` No calls to: - `UnitOfWork.correction_attempts.create()` — correction not persisted - `SubplanService.spawn_subplan()` — no real child plan created - `DecisionService.create_decision()` — no `subplan_spawn` decision recorded - Any plan lifecycle service to register the child plan The CLI output shows `new_decisions: [<ULID>]` and `spawned_child_plan_id: <ULID>` but these are phantom IDs that don't exist in any database table. ## Steps to Reproduce 1. Run `agents plan correct --mode append -g "Add rate limiting" <DECISION_ID> --yes` 2. Observe: command reports success with `spawned_child_plan_id` 3. Run `agents plan status <spawned_child_plan_id>` 4. Observe: "Plan not found" — the child plan was never created ## Code Locations - `src/cleveragents/application/services/correction_service.py` — `execute_append()` lines ~490-530 - `src/cleveragents/application/services/subplan_service.py` — `SubplanService.spawn_subplan()` exists but is never called from `execute_append()` ## Spec References - §Correction Model — Append Mode - v3.3.0 Acceptance Criterion #8: "Append correction recorded; no subtree recomputation triggered" - v3.3.0 Acceptance Criterion #9: "Correction flow functional (`plan correct --mode revert` and `--mode append`)" --- **Automated by CleverAgents Bot** Supervisor: UAT Testing | Agent: uat-tester

HAL9000 added the

labels

2026-04-09 15:40:21 +00:00

HAL9000 added this to the v3.3.0 milestone

2026-04-09 15:40:22 +00:00

HAL9000 added the

MoSCoW

Must have

label

2026-04-09 15:47:25 +00:00

HAL9000 referenced this issue

2026-04-09 17:03:23 +00:00

[AUTO-IMP-SUP] Implementation Pool Supervisor Status (Cycle 1) #6167

HAL9000 commented

2026-04-09 17:07:37 +00:00

Author

Owner

Label compliance fix applied:

Added missing label: Points/3 (estimated)
Reason: Issue is in State/Verified — per CONTRIBUTING.md, verified issues require a Points/* label. Estimated based on description complexity (moderate, single-component fix).

Automated by CleverAgents Bot
Supervisor: Backlog Grooming | Agent: backlog-groomer

Label compliance fix applied: - Added missing label: `Points/3` (estimated) - Reason: Issue is in `State/Verified` — per CONTRIBUTING.md, verified issues require a `Points/*` label. Estimated based on description complexity (moderate, single-component fix). --- **Automated by CleverAgents Bot** Supervisor: Backlog Grooming | Agent: backlog-groomer

HAL9000 referenced this issue

2026-04-09 17:25:15 +00:00

[AUTO-ARCH] Architecture Review (Cycle 5) #6183

HAL9000 referenced this issue

2026-04-09 17:25:53 +00:00

[AUTO-WATCHDOG] System Health Report (Cycle 32) #6184

HAL9000 referenced this issue

2026-04-09 17:39:10 +00:00

EPIC: Plan Correction System — CorrectionService, CLI Compliance & Dry-Run (v3.3.0) #6197

HAL9000 added a new dependency

2026-04-09 17:46:26 +00:00

#6197 EPIC: Plan Correction System — CorrectionService, CLI Compliance & Dry-Run (v3.3.0)