JOBNIMBUS MCP TOOL – BUG REPORT - 18102025-03.txt•5.01 kB
Aquí tienes el **texto plano profesional y técnico**, listo para enviar al desarrollador de la **MCP Tool JobNimbus (instancia Stamford)**, con los errores detectados, pasos de corrección y validaciones necesarias:
---
### 🧩 JOBNIMBUS MCP TOOL – POST-FIX VALIDATION REPORT (STAMFORD)
#### 1. **Issue: “Unassigned” task ownership still persists**
**Symptoms:**
All 420 tasks are grouped under `assignee_name: "Unassigned"` in the `assignment_analytics` block, even though the raw task data and `overdue_analysis` correctly show owners such as “Ana Macassi”, “Jeison Castro”, “Juan Villavicencio”, “Diana Castro”, and “Automation (Job)”.
**Root cause:**
The fallback mapping from `owner_id` to `created_by_name` (or `last_modified_by`) is being applied **after** aggregation instead of before.
**Required Fix:**
Move the fallback logic **before building the `assignment_analytics` object.**
```js
if (!assignee_name || assignee_name === 'Unassigned') {
assignee_name = created_by_name || last_modified_by || 'Unassigned';
}
```
* Apply this during task normalization, **before grouping** by assignee.
* Ensure this mapping also cascades into `productivity_score` calculations.
**Expected Outcome:**
Assignment distribution should show multiple team members, not a single “Unassigned” category.
---
#### 2. **Issue: Negative average completion time**
**Symptoms:**
`avg_completion_time_hours = -487845.0` hours (≈ –20,327 days) appears in both summary and task_type_metrics.
**Root cause:**
Incorrect date subtraction order or mismatched timezone conversion between `created_at` and `completed_at`.
**Required Fix:**
Reverse and validate timestamp subtraction:
```js
const diff = new Date(completed_at) - new Date(created_at);
avg_completion_time_hours = diff > 0
? diff / 3600000
: Math.abs(diff) / 3600000;
```
* Also ensure both timestamps are parsed in **UTC** or normalized to the same timezone before calculating differences.
**Expected Outcome:**
Average completion time should fall between **6–36 hours**, depending on task type.
---
#### 3. **Issue: Productivity trends static**
**Symptoms:**
All four weeks in `productivity_trends` show `0 tasks created / completed`, “Stable” trend.
Yet analytics show 312 completed tasks in total.
**Root cause:**
The weekly trend module is referencing filtered time windows with no overlap between creation and completion timestamps.
**Required Fix:**
Update time-window segmentation logic:
```js
group_by_week = task => getISOWeek(new Date(task.completed_at || task.created_at));
```
* Include tasks that started in one week and finished in another.
* Allow flexible window overlap for ongoing tasks.
**Expected Outcome:**
Weekly trends should show variations in completion counts (approx. 70–90 per week) and progressive improvement curves.
---
### ✅ Validation Plan
After applying the above fixes, the developer must run the following tests:
#### A. **Owner Mapping Validation**
```bash
get_task_management_analytics --days_back 60
```
**Expected:**
* `assignment_analytics` contains ≥ 5 different `assignee_name`.
* `"Unassigned"` represents <10% of total tasks.
* `workload_status` per user shows correct proportional distribution.
#### B. **Completion Time Validation**
Query average completion hours directly:
```bash
SELECT AVG(TIMESTAMPDIFF(HOUR, created_at, completed_at)) FROM tasks WHERE is_completed = true;
```
**Expected:** 6 ≤ avg ≤ 36 hours.
No negative values.
#### C. **Weekly Trend Validation**
```bash
get_task_management_analytics --include_productivity_trends true
```
**Expected:**
* Each week shows tasks_created > 0 and tasks_completed > 0.
* At least one week has a completion_rate > 70%.
#### D. **Cross-Check User Productivity**
```bash
get_user_productivity_analytics --days_back 60
```
**Expected:**
* `active_members > 4`
* `avg_productivity_score > 0`
* Top performer identified (not “N/A”)
---
### 🔍 Expected Final Output After Fix
| Metric | Expected Value | Status Goal |
| ------------------- | ----------------- | ----------- |
| Completion Rate | 70–80% | ✅ |
| Overdue Tasks | <15% | ✅ |
| Avg Completion Time | 6–36 h | ✅ |
| Distinct Assignees | ≥ 5 | ✅ |
| Unassigned % | <10% | ✅ |
| Weekly Trend | Active (non-zero) | ✅ |
---
### 💡 Summary for Developer
The core logic and API connection now function correctly.
Only the **post-processing layer** needs adjustment:
1. Apply `assignee` mapping before analytics aggregation.
2. Correct date-difference direction and timezone normalization.
3. Enable dynamic weekly trend segmentation.
Once fixed and validated, the MCP analytics engine will fully represent **real operational data** and can safely power Notion, Power BI, or GA dashboards without distortion.
