The /fdic-failure-forensics command reconstructs the pre-failure financial timeline for a single failed FDIC-insured institution, identifies the earliest visible warning signals from public data, and explains likely drivers of deterioration.

This is a Claude Code skill, not an MCP tool. It requires Claude Code with the plugin installed. If you are using another MCP client, you can approximate this workflow by combining fdic_get_institution_failure, fdic_search_financials, fdic_detect_risk_signals, and fdic_analyze_bank_health in sequence — see Choose a Workflow.

When to Use It

• You want to study a specific bank failure for training, pattern recognition, or case-study purposes.
• You need to understand what public data showed in the quarters leading up to a failure.
• You want to identify which risk signals appeared earliest and trace the deterioration path.

When Not to Use It

• The institution is still active: Use Bank Deep Dive instead. Failure Forensics requires a confirmed failure record.
• Screening a portfolio for current risks: Use Portfolio Surveillance instead.
• Quick failure lookup: If you only need the failure date, cost, and resolution type, use fdic_get_institution_failure directly.

Inputs

Input	Required	Description
Institution identity	Yes	Failed bank name or FDIC CERT number
Pre-failure report date	No	Defaults to the last quarter-end before the failure date
Lookback window	No	Number of quarters to analyze. Default 8.
Focus area	No	funding, credit, earnings, or overall (default). Determines which domain-specific tools are invoked.

Examples

/fdic-failure-forensics Silicon Valley Bank

/fdic-failure-forensics 24735

/fdic-failure-forensics First Republic Bank, lookback 12 quarters, focus on funding

/fdic-failure-forensics Heartland Tri-State Bank, lookback 8 quarters

What Output to Expect

A structured report with core sections that are always present and enrichment sections included when the data warrants them:

Section	Always Present	Contents
1. Institution Identification	Yes	Name, CERT, location, charter class, regulator
2. Failure Event Summary	Yes	Failure date, resolution type, estimated DIF cost, acquiring institution
3. Pre-Failure Financial Timeline	Yes	Quarter-by-quarter Call Report data over the lookback window
4. Earliest Warning Signals	Yes	Risk signals from the last reported quarter, with first-appearance timing
5. Likely Failure Drivers	Yes	Analytical narrative with every statement tagged as [Observed], [Inferred], or [Unknown]
6. Domain Analysis	When relevant	Funding profile or credit concentration, invoked only when deterioration implicates that domain
7. Regional Context	When available	State unemployment, rate environment, economic backdrop during the pre-failure period
8. Caveats / Limits of Public Data	Yes	Data gaps, temporal lag between last report and failure, what public data cannot observe

Provenance Tags

The Likely Failure Drivers section tags every analytical statement:

• [Observed]: Directly visible in public FDIC data (e.g., declining capital ratios, rising noncurrent loans)
• [Inferred]: Reasonable conclusion drawn from observed patterns (e.g., likely funding pressure based on deposit outflows and rising brokered deposits)
• [Unknown]: Cannot be determined from public data (e.g., liquidity run timing, off-balance-sheet exposures, confidential supervisory actions)

Key Caveats

• Failed institutions only: The skill requires a confirmed FDIC failure record. It will not run for active institutions.
• Temporal gap: There is always a gap between the last quarterly Call Report and the actual failure date. Events in that gap (bank runs, emergency actions) are not captured in the financial timeline.
• Quarterly data basis: The financial timeline uses quarterly Call Report data (REPDTE). Dollar amounts are in thousands.
• Publication lag: Some data may have been published after the failure, but it reflects the institution’s reported position at quarter-end.
• Not a complete post-mortem: Public data cannot capture liquidity runs, off-balance-sheet exposures, market sentiment, fraud, or confidential supervisory findings. The caveats section makes these limitations explicit.
• Proxy, not regulatory: Health and risk assessments are public-data analytical proxies.

Under the Hood

The skill orchestrates these MCP tools:

Tool	Purpose
fdic_search_institutions	Resolve institution identity and confirm CERT
fdic_get_institution_failure	Retrieve failure record (date, resolution type, cost, acquirer)
fdic_search_financials	Build quarter-by-quarter pre-failure financial timeline
fdic_detect_risk_signals	Surface risk signals visible at the last reported quarter
fdic_analyze_bank_health	CAMELS-proxy assessment for the last reported quarter
fdic_search_history	Structural events (mergers, charter changes) in the lookback window
fdic_analyze_funding_profile	Funding composition when deterioration implicates funding stress
fdic_analyze_credit_concentration	Credit concentration when deterioration implicates credit risk
fdic_regional_context	Macro/regional economic backdrop during the pre-failure period

Hard-dependency tools (institution search, failure record, financials, risk signals) must succeed or the skill stops. Soft-dependency tools (health assessment, structural history, regional context) degrade gracefully. Context tools (funding profile, credit concentration) are invoked only when the deterioration pattern implicates those domains.