Purpose
Systematically scan a single credit exposure against the full set of early warning indicators, assign a quantified deterioration score, and output a clear watchlist recommendation with a prescribed monitoring cadence and action plan — so emerging credit problems are caught before they become non-performing.
When to Use
Invoke this skill when the user asks to:
- Assess whether a specific borrower or credit exposure is showing early signs of deterioration
- Build or update a credit watchlist for portfolio review
- Decide whether to downgrade a credit rating or move an account to special mention/watch
- Justify increased monitoring frequency or a covenant review on an account
- Prepare an early warning summary for a credit committee or portfolio review meeting
- Investigate why an account's risk profile feels off despite a current pass rating
Inputs Required
Before starting, collect:
- Borrower name and exposure type: Entity and facility type (term loan, revolver, CRE, trade finance, etc.)
- Current rating and exposure size: Internal risk grade and outstanding/committed amount
- Recent financials: Last 2–3 periods — revenue, EBITDA, leverage, liquidity, DSCR/ICR if available
- Payment and behavioral history: Days past due, utilization trends, overdraft/limit breaches, drawdown patterns
- Covenant and collateral status: Active covenants, headroom, any waivers, collateral value/LTV
- Industry and market context: Sector, geography, known sector stress
- Qualitative inputs: Management changes, auditor changes, news, litigation, customer/supplier concentration
If financials are limited to one period or some inputs are missing, score on available data and flag each indicator as "insufficient data" rather than assuming a clean signal.
Instructions
Scan 6 Signal Categories and Score Each Indicator
For each indicator below, assign a status:
- Green (0 pts): No concern / stable or improving
- Amber (1 pt): Early-stage signal or negative trend within tolerance
- Red (2 pts): Material deterioration or threshold breach
- N/D: Insufficient data — flag, do not score as green
For every Amber or Red, cite the specific data point that triggered it.
Category 1 — Financial Deterioration (financial statements)
- Leverage trend: Net debt/EBITDA rising; Red if breaching the level implied by current rating
- Profitability: EBITDA margin compression over 2+ periods; revenue decline
- Debt service coverage: DSCR or ICR falling; Red if approaching 1.0x or below covenant
- Liquidity: Cash + undrawn lines vs. near-term maturities; current ratio deterioration
- Working capital: Lengthening DSO, stretching DPO (supplier strain), rising inventory days
Category 2 — Behavioral / Payment Signals (account activity)
- Days past due: Any movement from 0 → 1–29 → 30+ DPD
- Facility utilization: Revolver/overdraft drawn to or above limit, or sharp utilization spike
- Limit breaches / returned items: Excesses, bounced payments, repeated technical overdrafts
- Cash flow patterns: Declining or erratic credit turnover through operating accounts
- Refinancing behavior: Requests to extend, restructure, or roll short-term facilities
Category 3 — Covenant & Collateral Stress (loan documentation)
- Covenant headroom: Headroom shrinking; Red if breached or waiver requested/granted
- Waivers and amendments: Frequency increasing — pattern of repeated amendments
- Collateral value / LTV: Collateral coverage declining; Red if LTV breaches threshold
- Security perfection: Lapses, releases, or gaps in security position
- Guarantor strength: Deterioration in guarantor financials or support capacity
Category 4 — Market & Industry Pressure (external context)
- Sector outlook: Industry in structural or cyclical decline
- Peer comparison: Borrower underperforming sector peers on key metrics
- Input/cost exposure: Rising input costs, FX exposure, rate sensitivity on floating debt
- Demand signals: Order book, backlog, or end-market demand softening
- Regulatory/policy: Pending regulation or policy shifts threatening the business model
Category 5 — Qualitative & Management Red Flags (soft signals)
- Management/ownership change: Unexpected CEO/CFO departure, ownership change
- Auditor signals: Auditor change, going-concern note, qualified opinion, delayed accounts
- Reporting quality: Late, incomplete, or inconsistent financial submissions
- Concentration: Loss of major customer/supplier; high single-name concentration
- Reputation/legal: Adverse news, litigation, regulatory action, fraud indicators
Category 6 — Relationship & Disclosure Behavior (interaction signals)
- Responsiveness: Borrower slow or evasive on information requests
- Forward visibility: Reluctance to share forecasts, budgets, or updated projections
- Tone shift: Defensiveness, surprise requests, changed point of contact
- Banking relationships: New lenders entering, reducing facilities with the bank, deposit outflows
Score Aggregation and Watchlist Recommendation
- Sum the points across all scored indicators and compute a deterioration score.
- Map to a tier: - Stable (Performing): Low total, no Reds — maintain standard review - Monitor (Special Mention): Multiple Ambers or 1 isolated Red — increased monitoring - Watchlist: Several Ambers/Reds clustered, or any Red in covenant/payment categories — formal watchlist - Action Required (Substandard risk): Multiple Reds, especially across financial + covenant + behavioral — escalate immediately
- Identify the dominant driver: Name the single category contributing most to deterioration.
- Note correlated signals: Flag where multiple categories point to the same root cause (e.g., liquidity strain showing in both utilization spike and DPO stretching).
Prescribe Monitoring Cadence and Action Plan
Based on the tier, specify:
- Review frequency: Quarterly / monthly / weekly
- Information to request: Specific items (interim financials, 13-week cash flow, updated collateral valuation)
- Top 3 actions: Concrete next steps in priority order (e.g., request covenant compliance certificate, schedule borrower call, commission updated collateral valuation)
- Escalation trigger: The specific event that would force a downgrade or remedial referral
Output Format
- Structure: Header summary block (borrower, exposure, current rating, computed tier, score) → 6 scored category tables → aggregation and dominant driver → monitoring and action plan
- Length: 800–1,200 words
- Format: Scored tables per category (indicator / status / triggering data point), prose for aggregation and action plan
- Sections: Summary block → Category 1–6 scorecards → Score aggregation & watchlist recommendation → Monitoring cadence & action plan → Escalation trigger
- Tone: Credit risk officer — evidence-based, conservative, action-oriented; never assumes a clean signal where data is missing
Quality Checklist
Before delivering output, verify:
- [ ] All 6 categories are scored, with every indicator marked Green / Amber / Red / N/D
- [ ] Every Amber and Red cites the specific data point that triggered it — no unsupported flags
- [ ] Missing data is flagged as N/D, never silently scored as Green
- [ ] The deterioration score maps to one explicit watchlist tier
- [ ] The dominant driver category is named
- [ ] The action plan includes a review frequency, top 3 actions, and a specific escalation trigger
Notes
- Payment/behavioral and covenant signals are leading indicators; financial statement deterioration often lags by a quarter or more — weight a single Red in payment or covenant categories more heavily than a comparable Amber in financials.
- Clustered correlated signals across categories are far more predictive of default than scattered isolated Ambers — call this out explicitly.
- For thin-file or trade-finance exposures, behavioral and relationship signals (Categories 2 and 6) often carry more diagnostic value than financials.
- This assesses one exposure at a time; run repeatedly across a portfolio to build a ranked watchlist.
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills