aitask-stats

Usage

Run the statistics script:

./.aitask-scripts/aitask_stats.sh [OPTIONS]

Options

• -d, --days N - Show daily breakdown for last N days (default: 7)
• -w, --week-start DAY - First day of week: mon, sun, tue, etc. (default: Monday)
• -v, --verbose - Show individual task IDs in daily breakdown
• --csv [FILE] - Export raw data to CSV (default: aitask_stats.csv)
• --backlog-weeks N - Weeks of backlog history to render (default: 8, max: 99)
• --csv-backlog [FILE] - Export the weekly backlog series to CSV (default: aitask_backlog.csv)
• -h, --help - Show usage information

Examples

Basic statistics (last 7 days):

./.aitask-scripts/aitask_stats.sh

Extended daily view (14 days):

./.aitask-scripts/aitask_stats.sh --days 14

Verbose output with task names:

./.aitask-scripts/aitask_stats.sh -v

Export to CSV for graphing in LibreOffice:

./.aitask-scripts/aitask_stats.sh --csv

Statistics Provided

1. Summary - Total completions, 7-day and 30-day counts
2. Backlog Level - Weekly count of open tasks per category, with follow-up / genuine subtotals, TOTAL OPEN, and a parent/child split
3. Backlog Net Flow - Weekly arrivals minus departures per category, with ARRIVALS / DEPARTURES / NET rows
4. Daily Breakdown - Completions per day with optional task IDs
5. Day of Week Stats - Current week counts + 30d/all-time averages per weekday
6. Pipeline Timing - Average time in the implement and review/merge phases, for gated tasks
7. Label Weekly Trends - Per-label completions for last 4 weeks (W-3, W-2, W-1, This Week)
8. Label Day-of-Week Breakdown - Per-label averages by day of week
9. Task Type Weekly Trends - Parent/child and feature/bug trends for last 4 weeks
10. Features/Bugs by Label Trends - Combined label + issue type weekly trends
11. Code Agent Weekly Trends - Weekly completions split by code agent
12. LLM Model Weekly Trends - Weekly completions split by normalized LLM model
13. Verified Model Rankings - Model scores per skill, aggregated across providers

The backlog sections cover the last 8 weeks by default (--backlog-weeks changes the horizon). Their columns run chronologically with the current week last; in the net-flow table that final column is a partial week, marked Now*.

Export Format

CSV Export (--csv): one row per completed task, with columns:

• date, day_of_week, week_offset, task_id, labels, issue_type, task_type, implemented_with, codeagent, llm_model, created_at, category

Backlog Export (--csv-backlog): the weekly backlog series, with columns:

• week_ending, category, open, arrived, departed, net

One row per category per week, oldest week first, zero cells included. category holds raw namespaced keys (type:feature, kind:manual_verification) and only real categories -- no subtotal or TOTAL OPEN rows for a pivot to double-count. Open tasks are not rows in the --csv table, so the backlog level is not reproducible from it.

Open in LibreOffice Calc to create custom charts and pivot tables for trend analysis.

Importing CSV in LibreOffice Calc

1. Open LibreOffice Calc
2. File -> Open -> Select the CSV file
3. In the import dialog:
   • Character set: UTF-8
   • Separator: Comma
   • Check "Quoted field as text"
4. Click OK

Creating Charts

1. Select the data range
2. Insert -> Chart
3. Choose chart type (Line, Bar, or XY Scatter for trends)
4. Follow the wizard to customize

Pivot Tables for Analysis

1. Select all data
2. Insert -> Pivot Table
3. Drag fields:
   • Row: week_offset or day_of_week
   • Column: labels or issue_type
   • Data: Count of task_id
4. Creates summary table for trends