post-loan-management

Target Role

• Role: Post-loan Management Officer / Risk Reviewer
• Usage Scenario: Continuous post-loan management after loan disbursement - initial inspection, regular inspection, risk early warning, classification adjustment
• Output Purpose: Generate structured post-loan inspection reports as the basis for risk classification adjustment, early warning disposal, and regulatory inspection
• Decision Level: Provide risk analysis and classification suggestions; downgrading to non-performing category requires manual approval
• Execution Frequency: Determined based on customer risk classification and credit amount matrix (see references/check-frequency-policy.md)

Data Sources

Required Data

Data Desensitization Rules

• Unified Social Credit Code of Enterprise: Display first 6 and last 4 digits, replace the middle with

  *

• Customer Contact Phone Number: Only display first 3 and last 4 digits
• Bank Account Number: Only display last 4 digits
• Detailed Collateral Address: Do not display completely in output, only show region and type

Degradation Strategy

• If credit report data is unavailable: Mark "Not included in credit dimension", continue with other analysis
• If only 1 year of financial statements is available: Mark "Insufficient data, trend analysis not available", only conduct static analysis
• If fund flow details cannot be obtained: Mark "Fund flow verification cannot be executed, manual retrieval is recommended", continue with other inspections
• If collateral value data is unavailable: Mark "Collateral value is based on the latest assessment, which may not reflect current market price"
• If external industry data is unavailable: Use benchmark data in references/ and mark "Based on historical benchmark data"

Terminology

Workflow

Read Before Write: Before starting any analysis, the following data confirmation steps must be executed:

1. Read and list all input data (customer information, credit ledger, financial statements, credit report, etc.)
2. Confirm the time range, accounting standards (CAS/IFRS), and currency of the data
3. Run

   scripts/validate_post_loan_data.py

   to verify data integrity and consistency
4. Proceed to Step 1 only after verification passes

Step 1: Post-loan Inspection Planning and Preparation

1. Refer to

   references/check-frequency-policy.md

   to determine the inspection frequency matrix
2. Retrieve recent customer files, historical post-loan inspection records, and early warning disposal records
3. Query the customer's latest credit report, judicial information, and public opinion dynamics
4. Mark unresolved issues found in the previous period as key points for this verification

• ✅ Inspection plan generated successfully → Proceed to Step 2
• ❌ Basic customer information missing → Stop inspection, output missing list
• ⚠️ Partial data unavailable → Process according to data source degradation strategy, continue and mark in the report

📋 Data Source:

user_upload

(customer provides inspection parameters)

Step 2: Fund Usage and Flow Verification

references/fund-usage-policy.md

1. Verify entrusted payment fund flow: Check the authenticity of payee and transaction background against contract terms
2. Spot-check large autonomous payments (single transaction > 1 million yuan): Verify if the usage complies with contract terms
3. Execute fund return verification: Identify abnormal transactions returning to the borrower/related parties within 7/30 days
4. Verify capital adequacy and project progress matching for fixed asset loans
5. Mark suspected illegal flows item by item against the prohibited flow list

scripts/check_fund_usage.py --input {fund_flow_data} --rules references/fund-usage-policy.md

• ✅ No violations found → Proceed to Step 3
• ❌ Fund embezzlement found → Mark as a veto condition (P1), output red warning, proceed to Step 5 but highlight in red in the final report
• ⚠️ Suspicious transaction background → Mark as yellow warning, proceed to Step 3 and mark in the report

📋 Data Source:

system_api

(fund flow data from credit system)

Step 3: Business Status and Financial Health Inspection

1. Read financial statements and verify business status from the following dimensions:
   • Revenue changes (YoY/ MoM)
   • Orders and production (outstanding orders, capacity utilization)
   • Changes in major customers/suppliers
   • Inventory and turnover
   • Personnel changes
2. Calculate financial health indicators (current ratio, quick ratio, asset-liability ratio, interest coverage ratio, operating cash flow/total liabilities, accounts receivable turnover days), refer to

   references/industry-benchmarks.md

   for industry benchmarks
3. Determine if early warning is triggered against

   references/financial-warning-thresholds.md

4. Execute financial fraud identification checks (cross-validation of statements with tax control/flow, matching of revenue and cash flow, related party transactions, year-end突击回款)

• Do not skip any indicators, even if some indicators "seem normal"
• Show calculation process for all ratios, do not directly give conclusions
• If indicators do not meet expectations, stop to analyze the reasons

scripts/calculate_financial_ratios.py --input {financial_data} --benchmarks references/industry-benchmarks.md

• ✅ Normal business operation, financial indicators within normal range → Proceed to Step 4
• ⚠️ Partial indicators enter attention range → Mark warning signals, proceed to Step 4
• ❌ Multiple indicators enter warning range or financial fraud found → Mark orange/red warning, proceed to Step 4

📋 Data Source:

user_upload

(financial statements uploaded by user)

Step 4: Guarantee Validity and External Risk Environment Inspection

1. Verify one by one according to guarantee type (mortgage/pledge/guarantee):
   • Market value changes and physical status of mortgaged assets
   • Credit status of payment obligor of pledged assets
   • Financial and credit status of guarantor
2. Refer to

   references/collateral-policy.md

   for upper limit of mortgage ratio and warning thresholds for various guarantees
3. Evaluate changes in industry policies, regional risks, market risks, and credit environment
4. Quantify the impact of external environment changes on customer's repayment ability

• ✅ Valid guarantee, no major adverse changes in external environment → Proceed to Step 5
• ⚠️ Collateral value drops ≥ 15% but < 25% → Mark yellow warning, proceed to Step 5
• ❌ Collateral seized or guarantor's credit status seriously deteriorated → Mark red warning, proceed to Step 5

📋 Data Source:

system_api

(collateral management system, guarantor credit report)

Step 5: Early Warning Identification and Risk Classification Assessment

1. Summarize all warning signals identified in Steps 2-4, determine the level (yellow/orange/red) against

   references/early-warning-indicators.md

2. Refer to

   references/five-classification-policy.md

   (five-level classification standards) to assess whether the current classification is accurate
3. Make comprehensive judgment based on the principle of "substance over form" (overdue days are only for reference)
4. If classification adjustment is needed, attach detailed analysis basis

references/p1-p6-veto-conditions.md

scripts/evaluate_risk_classification.py --input {assessment_data} --policy references/five-classification-policy.md

• ✅ Accurate classification, no new major risks → Proceed to Step 6
• ⚠️ Classification needs downgrade but not non-performing → Output downgrade suggestion and basis, proceed to Step 6
• ❌ Classification needs downgrade to non-performing or veto condition triggered → Output red warning and classification downgrade suggestion, proceed to Step 6

📋 Execution Subject:

ai→human

(AI generates classification suggestions → Risk reviewer confirms) 📋 Confirmation Mechanism:

approve

(downgrade to non-performing category requires manual approval)

Step 6: Early Warning Disposal and Post-loan Report Generation

1. Verify warning signals (exclude false positives), assess risk severity
2. Formulate differentiated disposal plans based on warning levels:
   • Yellow warning: Increase monitoring frequency, require customer to supplement materials
   • Orange warning: Reduce exposure, add guarantees, adjust credit terms
   • Red warning: Initiate early loan collection, litigation preservation, transfer of non-performing loans
3. Refer to

   references/disposal-escalation-policy.md

   to determine reporting path and time limit
4. Generate post-loan inspection report using

   assets/post-loan-report-template.md

5. Attach disclaimer at the end of the report (use

   assets/disclaimer-template.md

   )

📋 Execution Subject:

ai→human

(AI generates report → Dual signature and filing by inspector and responsible person) 📋 Confirmation Mechanism:

approve

(report can be filed only after dual signature)

Output Format

assets/post-loan-report-template.md

1. Basic Customer Information

2. Overview of Current Inspection

3. Fund Usage Verification Results

4. Business and Financial Assessment

5. Guarantee Validity Evaluation

6. Early Warning Signal List

7. Risk Classification Suggestion

8. Disposal Suggestions and Next Inspection Plan

9. Disclaimer

assets/disclaimer-template.md

This output can be parsed and used by credit-risk-classification and early-warning-disposal Skills.

Compliance Constraints

1. Prohibit Profit Commitments: Under no circumstances shall there be deterministic statements such as "expected recovery", "expected improvement", "recovery rate is expected to be X%", etc.
2. Prohibit Data Guessing: Missing data must be requested from users or processed according to the degradation strategy. It is strictly prohibited to replace real data with industry averages (industry averages are only used for benchmarking comparison).
3. Data Timeliness Labeling: If the referenced industry benchmark data exceeds the marked validity period, it must be labeled in the output as "⚠️ Industry benchmark data may be outdated".
4. Prohibit Overstepping Approval Authority: This Skill only generates classification suggestions and disposal plans, and shall not replace manual approval or automatically execute classification adjustments.
5. Prohibit Concealing Risks: Do not downplay or omit identified warning signals; all warnings must be truthfully listed and graded.
6. Prohibit Bypassing On-site Inspection: Customers in Special Mention category or above must be inspected on-site; telephone or system verification shall not be used as a substitute.
7. Prohibit Post-hoc Backfilling: Post-loan inspection reports must be generated and filed in real time; post-hoc backfilling or tampering with historical records is prohibited.
8. Veto Conditions Trigger Immediate Reporting: If any of the P1-P6 conditions is triggered, red warning must be issued immediately and emergency reporting initiated without delay.

Audit Trail

audit/{Enterprise Abbreviation}_{Date}_post_loan_audit.json

{
  "skill_name": "post-loan-management",
  "skill_version": "1.0.0",
  "execution_time": "YYYY-MM-DDTHH:mm:ss+08:00",
  "customer_id": "[Desensitized]",
  "check_type": "Initial Inspection/Regular Inspection/Risk Classification Adjustment/Early Warning Disposal",
  "model": "claude-opus-4-7",
  "operator": "[Name] (Employee ID: [ID])",
  "steps": [
    {
      "step": "Data Confirmation and Verification",
      "executor": "ai",
      "data_source": {"type": "user_upload"},
      "result": "Pass/Fail",
      "duration_seconds": 0
    },
    {
      "step": "Fund Usage Verification",
      "executor": "ai",
      "data_source": {"type": "system_api", "system": "Credit System"},
      "result": "Pass/Fail/Warning",
      "findings_count": 0
    },
    {
      "step": "Business and Financial Inspection",
      "executor": "ai",
      "data_source": {"type": "user_upload"},
      "result": "pass/fail/warning"
    },
    {
      "step": "Guarantee Validity Inspection",
      "executor": "ai",
      "data_source": {"type": "system_api", "system": "Collateral Management System"},
      "result": "pass/fail/warning"
    },
    {
      "step": "Early Warning Identification and Classification Assessment",
      "executor": "ai→human",
      "data_source": {"type": "context"},
      "ai_output": "Suggested Classification: [Classification]",
      "confirmation": {"type": "approve", "approved_by": "[Name]", "role": "Risk Reviewer", "final_decision": "[Classification]"}
    },
    {
      "step": "Report Generation and Filing",
      "executor": "ai→human",
      "confirmation": {"type": "approve", "signed_by": "[Inspector] + [Responsible Person]"}
    }
  ],
  "warnings": ["If any"],
  "veto_conditions_triggered": [],
  "references_used": ["references/check-frequency-policy.md", "references/industry-benchmarks.md", "..."]
}

Gotchas

#1: Over-reliance on Overdue Days as the Sole Basis for Classification

• Symptom: Directly classify loans overdue for 90 days as substandard, ignoring the borrower's actual repayment ability and sufficient guarantee
• Cause: Over-reliance on overdue days, failure to execute comprehensive judgment based on "substance over form"
• Solution: Strictly follow Step 5 of the workflow - overdue days are only for reference; judgment must be made by synthesizing repayment ability, guarantee sufficiency, and recovery possibility

#2: Inadequate Fund Return Penetration Leading to Missed Judgments

• Symptom: Only verify direct payees, fail to identify funds returning to the borrower after multiple transfers
• Cause: Fund flow verification only stays at the first-level counterparty, fails to execute penetration analysis
• Solution: Follow Step 2 of the workflow to penetrate to the final payee, focus on abnormal paths where cumulative return exceeds 30% of the loan amount within 7/30 days

#3: Ignoring Over-guarantee by Guarantor

• Symptom: The total external guarantee of the guarantor has exceeded 50% of net assets, but it is still evaluated as "good"
• Cause: Only focus on the guarantor's own financial indicators, fail to verify its total external guarantee
• Solution: Follow Step 4 of the workflow, guarantor verification must include the ratio of external guarantee/net assets; ratio > 50% requires warning

#4: Outdated Industry Benchmark Data Leading to Misjudgment

• Symptom: Use outdated industry benchmark data, misclassify normally operating customers as financial warning
• Cause: The data validity period marked in references/industry-benchmarks.md has expired but not updated
• Solution: Check the data validity period of all references/ files before execution; expired data must be labeled "may be outdated" or rejected

Examples

Example 1: Regular Post-loan Inspection for Normal Category Customer

Please conduct this quarter's post-loan inspection for XX Technology Co., Ltd. Current classification: Normal, credit amount: 80 million yuan, guarantee method: real estate mortgage.

1. Data Confirmation: Read customer files, credit ledger, latest financial statements → Verify data integrity → Pass
2. Inspection Plan: Refer to check-frequency-policy.md → Normal category ≥ 50 million yuan → 1 on-site + system inspection per month
3. Fund Usage Verification: Verify entrusted payment vouchers and large autonomous payments → No violations found → Pass
4. Business and Financial Inspection: Calculate financial indicators → Current ratio 1.8, asset-liability ratio 52%, interest coverage ratio 4.2 → All within normal range → Pass
5. Guarantee Inspection: Mortgaged real estate value is stable, title certificate is valid → Pass
6. Early Warning Identification: No new warning signals → Maintain normal classification
7. Generate post-loan inspection report (including disclaimer) → Dual signature and filing

Example 2: Special Mention Category Customer Triggering Orange Warning

XX Manufacturing Co., Ltd. has been operating unstable recently, please conduct a post-loan inspection. Current classification: Special Mention, credit amount: 30 million yuan, guarantee method: guarantee.

1. Data Confirmation: Read data → Financial statements show revenue decreased by 25% YoY, operating cash flow has been negative for two consecutive quarters → Verification passed
2. Inspection Plan: Special Mention category → 1 on-site inspection per month
3. Fund Usage Verification: Found 1 autonomous payment of 2 million yuan transferred to a related enterprise, transaction background is suspicious → Mark yellow warning
4. Business and Financial Inspection: Revenue decreased by 25% (> 20% attention threshold), interest coverage ratio 1.2 (< 1.5 warning range) → Mark orange warning
5. Guarantee Inspection: Guarantor's asset-liability ratio is 68% (close to 70% threshold) → Mark attention
6. Early Warning Identification: Summarize 1 orange warning + 2 yellow warnings → Suggest reducing exposure and adding guarantees
7. Generate post-loan inspection report → Report to risk management department

Out of Scope

• This Skill does not perform pre-loan due diligence or credit approval (please use credit approval-related Skills)
• This Skill does not generate legal opinions or litigation strategies (please contact the legal department for litigation-related matters)
• This Skill does not directly execute fund transfer, early loan collection, litigation preservation, etc. (only generates suggestions)
• This Skill does not handle post-loan management of personal credit/retail business
• This Skill does not provide investment advice or asset disposal plans (please contact the asset preservation department for non-performing asset disposal)
• If users request the above content, clearly inform them and suggest contacting the corresponding department or using appropriate Skills