framework · cross business
Weekly maintenance checklist
Keep a private weekly record of failures, duplicates, approvals, source and permission changes, actual spend, reruns, owners, and the next review.
Purpose
Record a recurring maintenance decision with evidence, owners, and follow-up actions. Check the workflow’s results and boundaries, not merely whether its connections still respond.
When to use it
Use this checklist each week for the teaching pattern, after relevant changes, and after an incident. Urgent failures need action when discovered; do not wait for a scheduled review. Increase review frequency when workload or consequence requires it.
Reusable worksheet
Values stay in this page’s memory; refresh clears them. Nothing is submitted or saved automatically. Copy notes into your own document to keep them. Do not enter private records or credentials.
Filled synthetic example
| Check | Example, not an observed review |
|---|---|
| Workflow | Support drafts using FAQ F01 revision 1. |
| Failures | Hypothetical interrupted write; reconcile message M01 before replay. |
| Duplicates | Inspect ledger and review sheet for extra M01 effects. |
| Approvals | Pending drafts remain unsent; owner checks backlog age. |
| Source | Hypothetical guide location changed; withhold stale F01. |
| Permissions | Owner verifies connections remain restricted to approved sources. |
| Spend | Unknown; retrieve actual account charges and human time. |
| Rerun | Not run; test new FAQ and prior regression cases after correction. |
| Decision | Proposed pause until source and uncertain write are resolved. |
| Ownership | Assign actual owner, backup, and next review before operation. |
Instructions
Reconcile accepted inputs with completed, pending, rejected, and failed items. Inspect destination state for uncertain writes. Record identifiers and errors needed for recovery, avoiding unnecessary private contents in the maintenance log.
Review source revisions, revoked access, reviewer availability, and usage growth. Compare observed costs with the assumptions in the cost worksheet. Add representative failures and corrections to the test set. Record actual rerun IDs and verdicts rather than changing a tested date without execution.
Interpretation
Continue only within the accepted scope and review capacity. Stop on unexpected access, unauthorized effects, unresolved duplicate writes, unsupported commitments, or a backlog the owner cannot manage. Assign each unresolved item to a person and the manual fallback. Resume only after reconciliation, correction, and reviewed tests; preserve the history explaining the decision.
Limitations and sources
This is an original operating aid, not a monitoring service or guaranteed incident-detection method. The example is synthetic. A weekly checklist does not replace task-specific alerts, permission controls, or an owner available for urgent problems.
Related playbook and next step
Use it alongside the support playbook and maintenance guide. Practice the handoff with the backup. Suggest a correction.
Change notes
Foundation draft prepared October 20, 2026. No public publication or successful live-tool test is claimed.
Next learning step: Triage support emails and draft replies for review