UAT: `_run_estimation()` always uses `EstimationStubActor` instead of dispatching to the configured actor registry #3975

New issue

Open

opened 2026-04-06 08:09:35 +00:00 by freemo · 0 comments

freemo commented

2026-04-06 08:09:35 +00:00

Owner

Metadata

Branch: fix/estimation-real-actor-dispatch
Commit Message: fix(plan-lifecycle): dispatch estimation to actor registry instead of hardcoded stub
Milestone: None (Backlog)
Parent Epic: #890

Backlog note: This issue was discovered during autonomous operation
on milestone . It does not block milestone completion and has been
placed in the backlog for human review and future milestone assignment.

Bug Report

Feature Area: Estimation and Planning Intelligence
Severity: High — the estimation actor configuration is silently ignored at runtime
Found by: UAT tester (code-level analysis)

Background and Context

The spec defines an Estimation Actor as a configurable actor (set via --estimation-actor flag or action.estimation_actor YAML field) that analyzes plan complexity and produces an EstimationResult with cost, time, risk, and step estimates. The spec states:

"When enabled, a specialized estimation actor analyzes: the strategy produced by the Strategize phase..."

The Plan domain model correctly stores estimation_actor: str | None and estimation_result: EstimationResult | None. The use_action() method in PlanLifecycleService correctly resolves the estimation actor via a 4-level fallback chain (plan > action > global config).

Current Behavior

In PlanLifecycleService._run_estimation() (line 340-384 of src/cleveragents/application/services/plan_lifecycle_service.py), the implementation always instantiates EstimationStubActor regardless of what actor is configured on the plan:

# TODO: Replace EstimationStubActor with real actor dispatch
# via actor registry
stub = EstimationStubActor()
result = stub.estimate(plan.identity.plan_id)

The EstimationStubActor.estimate() method (in plan_executor.py) always returns:

EstimationResult(summary="Estimation skipped (stub actor)")

This means:

Any actor configured via --estimation-actor openai/gpt-4 is completely ignored
The estimation_result stored on the plan is always a placeholder with no real data
Users who configure an estimation actor get no actual estimation intelligence
The PLAN_ESTIMATION_COMPLETE event is emitted with a stub summary, misleading consumers

Expected Behavior (per spec)

When plan.estimation_actor is set to a real actor name (e.g., openai/gpt-4), the lifecycle service should:

Resolve the actor from the Actor Registry
Invoke the actor with the plan's description, definition of done, and decision tree
Parse the actor's response into an EstimationResult with real cost/time/risk estimates
Store the result on the plan

Steps to Reproduce

Create an action with estimation_actor: openai/gpt-4
Use the action: agents plan use my-action my-project
Execute the plan: agents plan execute <PLAN_ID>
Check the estimation result: agents plan status <PLAN_ID> --format json | jq .estimation
Observe: {"summary": "Estimation skipped (stub actor)"} — no real estimation

Code Location

File: src/cleveragents/application/services/plan_lifecycle_service.py
Method: _run_estimation() (lines ~322-384)
TODO comment: Line 343: # TODO: Replace EstimationStubActor with real actor dispatch

Subtasks

Implement real actor dispatch in _run_estimation() using the Actor Registry
Resolve the actor by name from plan.estimation_actor via unit_of_work.transaction()
Invoke the actor with plan context (description, definition_of_done, decision tree)
Parse the actor response into EstimationResult fields
Remove the EstimationStubActor usage from _run_estimation() (keep the class for testing)
Add unit tests (Behave feature file) for real actor dispatch path
Update integration tests to cover the estimation actor invocation

Definition of Done

_run_estimation() dispatches to the configured actor from the Actor Registry
EstimationStubActor is only used in tests, not in production code paths
A plan with estimation_actor set produces a real EstimationResult after execute_plan()
All existing estimation tests continue to pass
New tests cover the real actor dispatch path
PR merged and issue closed
All nox stages pass
Coverage >= 97%

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

## Metadata - **Branch**: `fix/estimation-real-actor-dispatch` - **Commit Message**: `fix(plan-lifecycle): dispatch estimation to actor registry instead of hardcoded stub` - **Milestone**: None (Backlog) - **Parent Epic**: #890 > **Backlog note:** This issue was discovered during autonomous operation > on milestone <M>. It does not block milestone completion and has been > placed in the backlog for human review and future milestone assignment. --- ## Bug Report **Feature Area:** Estimation and Planning Intelligence **Severity:** High — the estimation actor configuration is silently ignored at runtime **Found by:** UAT tester (code-level analysis) --- ## Background and Context The spec defines an **Estimation Actor** as a configurable actor (set via `--estimation-actor` flag or `action.estimation_actor` YAML field) that analyzes plan complexity and produces an `EstimationResult` with cost, time, risk, and step estimates. The spec states: > "When enabled, a specialized estimation actor analyzes: the strategy produced by the Strategize phase..." The `Plan` domain model correctly stores `estimation_actor: str | None` and `estimation_result: EstimationResult | None`. The `use_action()` method in `PlanLifecycleService` correctly resolves the estimation actor via a 4-level fallback chain (plan > action > global config). ## Current Behavior In `PlanLifecycleService._run_estimation()` (line 340-384 of `src/cleveragents/application/services/plan_lifecycle_service.py`), the implementation **always** instantiates `EstimationStubActor` regardless of what actor is configured on the plan: ```python # TODO: Replace EstimationStubActor with real actor dispatch # via actor registry stub = EstimationStubActor() result = stub.estimate(plan.identity.plan_id) ``` The `EstimationStubActor.estimate()` method (in `plan_executor.py`) always returns: ```python EstimationResult(summary="Estimation skipped (stub actor)") ``` This means: 1. Any actor configured via `--estimation-actor openai/gpt-4` is **completely ignored** 2. The `estimation_result` stored on the plan is always a placeholder with no real data 3. Users who configure an estimation actor get no actual estimation intelligence 4. The `PLAN_ESTIMATION_COMPLETE` event is emitted with a stub summary, misleading consumers ## Expected Behavior (per spec) When `plan.estimation_actor` is set to a real actor name (e.g., `openai/gpt-4`), the lifecycle service should: 1. Resolve the actor from the Actor Registry 2. Invoke the actor with the plan's description, definition of done, and decision tree 3. Parse the actor's response into an `EstimationResult` with real cost/time/risk estimates 4. Store the result on the plan ## Steps to Reproduce 1. Create an action with `estimation_actor: openai/gpt-4` 2. Use the action: `agents plan use my-action my-project` 3. Execute the plan: `agents plan execute <PLAN_ID>` 4. Check the estimation result: `agents plan status <PLAN_ID> --format json | jq .estimation` 5. Observe: `{"summary": "Estimation skipped (stub actor)"}` — no real estimation ## Code Location - **File**: `src/cleveragents/application/services/plan_lifecycle_service.py` - **Method**: `_run_estimation()` (lines ~322-384) - **TODO comment**: Line 343: `# TODO: Replace EstimationStubActor with real actor dispatch` ## Subtasks - [ ] Implement real actor dispatch in `_run_estimation()` using the Actor Registry - [ ] Resolve the actor by name from `plan.estimation_actor` via `unit_of_work.transaction()` - [ ] Invoke the actor with plan context (description, definition_of_done, decision tree) - [ ] Parse the actor response into `EstimationResult` fields - [ ] Remove the `EstimationStubActor` usage from `_run_estimation()` (keep the class for testing) - [ ] Add unit tests (Behave feature file) for real actor dispatch path - [ ] Update integration tests to cover the estimation actor invocation ## Definition of Done - [ ] `_run_estimation()` dispatches to the configured actor from the Actor Registry - [ ] `EstimationStubActor` is only used in tests, not in production code paths - [ ] A plan with `estimation_actor` set produces a real `EstimationResult` after `execute_plan()` - [ ] All existing estimation tests continue to pass - [ ] New tests cover the real actor dispatch path - [ ] PR merged and issue closed - All nox stages pass - Coverage >= 97% --- **Automated by CleverAgents Bot** Supervisor: UAT Testing | Agent: ca-new-issue-creator

freemo added a new dependency

2026-04-06 08:10:09 +00:00

#890 feat(actor): implement estimation actor as a functional actor type

freemo referenced this issue

2026-04-06 08:13:39 +00:00

[Automated] CleverAgents Build Session - 2026-04-06 #3775

freemo added the

Points

label

2026-04-06 08:15:05 +00:00

freemo referenced this issue

2026-04-06 08:23:38 +00:00