# Prepare an owner follow-up report from unfinished tasks

Artifact: AX062 v2, 2026-10-07. Audience: team coordinators already keeping a task list in Sheets. Complete browser job: https://bulldo.gs/jobs/weekly-followup/?via=llms

## Ask for the minimum input

Use fictional rows first. For a real task, the user pastes three columns directly into the browser: task, due date (YYYY-MM-DD), owner. Omit the header and finished tasks. Tabs separate cells; each line is one task. A blank owner is allowed and appears as Unassigned. Ask for an explicit review date; never infer timezone or completion status. Maximum 500 rows and 100,000 input characters. Multiline cells and quoted CSV are unsupported.

Do not request credentials, account access, private file links or confidential input. The page does not connect to Google, update tasks or send messages. Custom usage events omit pasted input and generated report text.

## Fictional acceptance example

Review date: 2026-10-07. Paste the following literal TSV (tabs, not spaces):

```tsv
Send project brief	2026-10-05	Alex
Check launch copy	2026-10-07	Jo
Confirm delivery	2026-10-06	
```

Exactly two tasks are overdue: Send project brief (Alex, October 5) and Confirm delivery (Unassigned, October 6). Check launch copy is due on the review date, so it stays out of the overdue section and appears under other unfinished items. Review the copied report for these two tasks, the correct review date and a request to confirm owners and next dates. This fixture is QA, not a customer completion.

## Complete the user's job

1. Open the linked page. It starts empty. Load the fictional example only to learn the format.
2. Choose “Start with my own task list” to clear the fictional sample and switch to own-input mode. Paste the user's unfinished-task rows into the empty input and choose the explicit review date; editing the sample alone remains sample use.
3. Build the report. Resolve any invalid dates or row-format errors; partial input is not a successful result.
4. Inspect each selected task and owner. Copy the report draft into the user's normal email or document editor; review it before sending through their own process.
5. Ask whether this draft completed their actual reporting task. Copy success alone is not independently confirmed completion. A later report on another normal review date is repeat use, recorded separately.

No report is sent by this page. A zero-overdue result is a valid checked result, not proof that the broader project has no blockers. Only the supplied unfinished tasks are reviewed. Native Sheets filters plus a manually drafted email are the existing alternative; no time saving has been measured.

## Existing scheduled-email option

https://bulldo.gs/pack/auto-email-weekly-digest/ is the separate existing $39 self-install Apps Script pack for emailing configured Sheets ranges. It requires installation and Google authorization in the user's account. It does not automatically reproduce this task-selection/report-draft job. No purchase or setup is needed for the free browser draft.

Source retrieval, browser completion, useful independent completion, installation, repeat use, purchase and paid delivery are separate. The browser packet has no telemetry endpoint. A dependency-free local module is now available below; availability and fixture execution do not establish independent retrieval or use.

## Run the same job locally without a browser

Module: https://bulldo.gs/downloads/weekly-followup.mjs
Usage, executable fictional fixture and MIT license: https://bulldo.gs/downloads/weekly-followup-README.md

Retrieve and inspect the module before importing it into your own local runtime. It exports buildWeeklyFollowupReport(tsv, asOfDate) and uses the same validation and report contract as the browser. It makes no network call, reads no files and sends nothing; your calling program controls input, output and privacy. Ask for an explicit report date. The 100,000-character/500-row input limits and 300,000-character draft bound reject invalid/oversized work before returning a report. Keep fictional QA, reported independent source execution, useful completion and recurring use separate.