The weekly defect report says the board is clean and the board is not

Problem Description

Every Monday a scheduled job posts a defect summary to the leadership channel. Last week it reported fifty open bugs and zero of the top urgency band, and on that basis the release was signed off. On Wednesday a customer escalation turned out to be one of six top-band tickets that were open the whole time.

I pulled the board by hand on the same morning the job ran; those counts are in the export below alongside the report the job produced and the script that produces it. The job has never errored - it posts a clean report every week and the person who wrote it has left.

I need to know how every wrong number in that report was produced, and I need it ordered by how badly each one misled the sign-off, because we have to tell the leadership channel something specific on Monday.

Output Specification

Produce exactly two files:

1. report-audit.md - each incorrect figure in the posted report, what the real figure is, the specific mechanism in the script that produced the wrong one, and how badly it misled the reader - ordered worst first. Include any defect in the script that has not yet produced a visibly wrong number but will. State plainly whether last week's report could have been trusted.
2. query-fixes.csv - columns figure,mechanism,fix,verification, where verification is the check that proves the figure is now complete.

Out of scope: redesigning the report format, changing what the leadership channel receives, and triaging any of the tickets themselves.

Input Files

Extract the following files before beginning.

=============== FILE: reports/2026-08-10-weekly.md ===============

Defect summary - week of 2026-08-10

• Open bugs: 50
• Top urgency (P1) open: 0
• Untriaged over 1 business day: 4
• Closed this week: 18

No top-urgency defects outstanding. Release sign-off recommended.

=============== FILE: exports/manual-board-count-2026-08-10.csv =============== figure,value,note open_bugs,137,"counted from the board on 2026-08-10 09:15" p1_open,6,"work items 8821, 8834, 8902, 8915, 8930, 8944" untriaged_over_1_day,41,"" closed_this_week,18,"matches the report" area_path_used_by_the_job,"Payments' EU","area path was renamed on 2026-07-20; the apostrophe is part of the name"

=============== FILE: automation/triage_report.py =============== import requests, os, base64

PAT = os.environ["ADO_PAT"] AUTH = base64.b64encode(f":{PAT}".encode()).decode() BASE = "https://dev.azure.com/acme/Payments" H = {"Authorization": f"Basic {AUTH}", "Content-Type": "application/json"} AREA = "Payments' EU"

def wiql(query, top=50): r = requests.post(f"{BASE}/_apis/wit/wiql?$top={top}&api-version=7.1", json={"query": query}, headers=H) if r.status_code != 200: return [] return r.json().get("workItems", [])

def open_bugs(): return wiql("SELECT [System.Id] FROM WorkItems " "WHERE [System.WorkItemType] = 'Bug' " "AND [System.State] NOT IN ('Resolved', 'Closed') " f"AND [System.AreaPath] = '{AREA}'")

def p1_open(): return [i for i in open_bugs() if field(i["id"], "Microsoft.VSTS.Common.Priority") == 1]

def field(work_item_id, name): r = requests.get(f"https://dev.azure.com/acme/_apis/wit/workitems/{work_item_id}" f"?$expand=all&api-version=7.1", headers=H) if r.status_code != 200: return None return r.json()["fields"].get(name)

if name == "main": opened = open_bugs() print(f"- Open bugs: {len(opened)}") print(f"- Top urgency (P1) open: {len(p1_open())}")