Download the complete part-03 practice pack (ZIP). Extract it and preserve its folders. All organizations, records, and results in this teaching pack are fictional.
These forms support Chapters 10–13. The bundled cases are fictional. Expected answers and dispositions are teaching criteria, not observed model results. No email, appointment, payment, policy change, or application permission change is authorized by completing an exercise.
Email and administrative queue
Case ID:
Message/thread reference and latest update time:
Current request:
Confirmed facts and evidence:
Missing information:
Owner and backup:
Status:
Next action and genuinely agreed due time:
Draft reference/version:
Recipient and attachment checks:
Decision or approval needed:
Approved final version:
Actual send/action confirmation:
Outcome and follow-up:
Use practice/email-cases.csv to distinguish a request, a verified fact, and a completed action. CASE-201's replacement appointment is unconfirmed. CASE-202 has no confirmed arrival date. CASE-203 requires the separate payment-change verification process.
Proposed meeting record
| Source note ID | Decision, action, proposal, or open question | Item | Owner if agreed | Due date if agreed | Review status |
|---|---|---|---|---|---|
| Complete from notes |
Read practice/Meeting-Notes.md, make your own record, and compare the three action items with practice/meeting-actions-expected.csv. Keep the discount proposal unapproved and the backup-coverage question unresolved.
Operating procedure template
Procedure ID and title:
Purpose and scope:
Status: draft / approved / superseded / withdrawn
Owner and approver:
Version and effective date:
Trigger:
Required inputs and access:
Responsible role and backup:
Steps:
Decision points and exceptions:
Actions requiring approval:
Evidence of completion:
Failure/manual fallback:
Practice or validation performed:
Unresolved issues:
Review triggers:
An effective date belongs to an approved procedure. A draft may have a proposed date, clearly labeled, but that does not establish that it is in force.
Document review record
Original document ID/location:
Source/received time:
Document type and version:
Related order, receipt, or case:
Fields extracted with source locations:
Unknown or ambiguous fields:
Arithmetic checks:
Cross-document comparisons:
Possible duplicates/revisions and processing history:
Exceptions:
Responsible reviewer:
Corrections and supporting evidence:
Approval status and approved version:
Actions taken and actual confirmations:
Retention/access requirements:
For INV-204, calculate the total and compare the billed quantities with PO-204 and REC-204. The arithmetic matches $152, but one B-20 belt is billed without a matching receipt quantity. The due date is unknown. The expected review result remains unresolved for an authorized person.
Internal knowledge service card
Service purpose and audience:
Covered questions:
Excluded uses and information:
Service owner and backup:
Approved source collection:
Source owners and approval states:
Role permissions:
How questions and feedback enter:
How answers show evidence and limits:
Escalation owner and response target:
How resolution is confirmed:
Source update/index refresh process:
Manual fallback:
Measures of quality, resolution, time, and cost:
Review triggers:
Knowledge issue queue
| Issue ID | Question or defect | Evidence/source | Category | Owner | Next action | Status | Resolution evidence |
|---|---|---|---|---|---|---|---|
| Complete for issue | Missing / wrong / outdated / conflicting / access |
Review practice/knowledge-cases.csv using practice/Knowledge-Sources.md. Two cases describe application access/update tests: they require a real authorized test environment and cannot be passed by a written answer alone.
Check the supplied example calculations
From the practice directory, with Python 3 installed, run:
python3 verify_examples.py
The script uses the Python standard library. It reads the fictional files and checks invoice arithmetic, receipt differences, action dates, expected statuses, and the chapter's workload calculations. It makes no network call and performs no business action. Its success establishes consistency of the teaching files, not the performance of an AI workflow.