---
name: cn-ib-diligence-workpaper
description: Structure Chinese investment-banking due-diligence materials into traceable schedules, testing workpapers, exception logs, and closure records. Use for IPO、再融资、并购重组 or sponsorship diligence involving 关联方、董监高、银行流水、资金往来、客户供应商、合同、资质证照、房产租赁、子公司、股权沿革、合规证明、抽样核查、访谈、函证 or Excel底稿. Use cn-ib-document-control when the task is only tracking receipt, signatures, originals, scans, and versions.
---

# CN IB Diligence Workpaper

Turn unstructured evidence into a reproducible workpaper that links objective, population, procedure, evidence, exception, conclusion, reviewer, and closure.

## Principles

1. Start from the diligence objective, not from the available files.
2. Preserve population, period, entity, sample method, and evidence provenance.
3. Separate factual evidence, management explanation, analyst calculation, and professional conclusion.
4. Never infer `no exception` from missing evidence.
5. Record contradictory evidence rather than selecting the more convenient source.
6. Do not claim sponsor, legal, or accounting sign-off. Route judgment to the responsible professional.
7. Apply confidentiality and personal-data minimization. Never place live bank-account or identity data in public examples.

## Workflow

### 1. Define the workpaper objective

State:

- assertion or risk addressed;
- entity and period;
- population and completeness source;
- materiality or review threshold;
- planned procedure;
- expected evidence;
- responsible preparer and reviewer.

Read `references/workstreams.md` for common workstreams. Tailor the objective to the transaction instead of copying a generic checklist.

### 2. Build the population

Before selecting samples or summarizing findings:

1. identify the source system or source schedule;
2. record extraction date and preparer;
3. reconcile population totals to an independent record where possible;
4. preserve excluded records and exclusion rationale;
5. mark incomplete populations as a scope limitation.

Do not treat a management-prepared list as complete without a completeness procedure.

### 3. Select and document procedures

Possible procedures include:

- inspection;
- recalculation;
- external confirmation;
- interview;
- database or public-record search;
- source-to-ledger tracing;
- ledger-to-source vouching;
- cross-document comparison;
- analytical review;
- full-population rule scan;
- risk-based or statistical sampling.

Record why the procedure addresses the stated risk. For samples, record population size, selection logic, sample size, and selected IDs.

### 4. Record evidence

Use `assets/workpaper-template.csv` or equivalent XLSX fields. Read `references/evidence-standard.md`.

Minimum lineage:

```text
workpaper_id, workstream, objective, procedure, subject, period,
population_source, sample_basis, source_document, source_location,
evidence_date, preparer, reviewer, status, exception,
management_explanation, conclusion, follow_up, closure_evidence
```

Retain calculations as formulas in spreadsheets. Keep assumptions in labeled cells and cite source documents in comments or evidence columns.

### 5. Evaluate exceptions

Read `references/exception-protocol.md`. For each exception:

- describe the observed fact;
- state the expected condition;
- quantify the amount or scope where possible;
- obtain and label management explanation;
- identify corroborating or contradictory evidence;
- assess whether the population or other samples may be affected;
- assign owner and deadline;
- record the closure test and evidence.

Avoid vague labels such as `已解释` without the explanation and corroboration.

### 6. Validate the workpaper

For a CSV-form workpaper, run:

```bash
python3 scripts/validate_workpaper.py workpaper.csv --output review-issues.csv
```

The script checks required lineage, duplicate IDs, invalid statuses, unsupported conclusions, and exception closure. Inspect the original evidence before resolving its findings.

For Word-format diligence questionnaires, run:

```bash
python3 scripts/scan_questionnaire_completeness.py questionnaire.docx --output questionnaire-review.csv
```

The scanner distinguishes unchecked yes/no controls from drafting placeholders. It flags conflicting or unanswered selections and affirmative answers with apparently blank detail fields. A signature warning means only that execution cannot be verified from the DOCX package; it is not proof that a signature or seal is missing. Convert legacy `.doc` files only in a temporary directory before scanning.

### 7. Form the conclusion

Conclude only within the tested scope. Use formulations such as:

- `在上述核查范围及程序下，未发现……`;
- `发现以下例外，尚待……`;
- `由于未取得……，暂无法就……形成结论`.

Do not use `不存在任何问题` when testing was sampled, evidence was incomplete, or exceptions remain open.

### 8. Review and close

Completion requires:

- objective and population are defined;
- source and location are traceable;
- calculations are reproducible;
- exceptions have owners and follow-up;
- conclusion matches evidence and scope;
- reviewer and review date are recorded;
- resolved exceptions retain closure evidence;
- confidential data is protected.

## Outputs

Deliver:

1. structured workpaper schedule;
2. exception log;
3. evidence-gap list;
4. open follow-up list by owner;
5. scoped conclusion and limitations;
6. reviewer-ready closure status.

Use temporary directories for extraction and previews. Save only final requested workpapers in the deal folder.