Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Financial Data Anomaly Detection & Investigation Engineer

    1

    Detect, investigate, and document financial anomalies, ledger errors, and policy exceptions with an audit-ready trail.

    $29.90

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Financial Data Anomaly Detection & Investigation Engineer

    Financial Data Anomaly Detection & Investigation Engineer

    Example session with this skill installed

    I have a CSV of 5,000 vendor payments from Q3. I need to find any duplicate payments or transactions that look like they were split to stay under our $5,000 approval threshold.

    • Read your context and instructions
    • Compiled the financial data anomaly

    I have profiled the vendor payment data for Q3. Total records: 5,000. Total value: $12,450,200. I have identified 12 high-priority exceptions across duplicate and threshold-avoidance categories.

    Financial Anomaly Register

    Case IDCategoryDetection ReasonEntity/VendorObserved ValueExposurePriorityConfidence
    ANOM-001DuplicateExact match: Vendor, Date, AmountTechFlow Inc.$4,850.00$4,850.00CriticalHigh
    ANOM-002Split3 payments to same vendor < 24hrsGlobal Logistics$14,200.00$14,200.00HighHigh
    ANOM-003ThresholdTransaction at $4,999.00Apex Consulting$4,999.00$4,999.00MediumMedium
    ANOM-004Near-DupSame vendor/amount; diff invoice #Office Supply Co$1,250.00$1,250.00MediumHigh

    Investigation Detail: ANOM-002 (Split Transaction)

    • Pattern: Three payments of $4,800, $4,750, and $4,650 issued to Global Logistics on August 14th.

    Context: The cumulative total of $14,200 exceeds the $5,000 individual approval limit. All three entries were posted by the same user ID (US-992) within a 4-hour window.

    • Evidence: Records 402, 405, and 410 share the same Purchase Order reference (PO-8821).

    Recommendation: Review with the Procurement Manager to determine if this was a legitimate staggered delivery or an intentional bypass of the authorization workflow.

    Next steps

    1. Validate the invoice images for ANOM-001 to confirm if the duplicate was a system error or double-billing.
    2. Request the approval logs for User ID US-992 regarding the Global Logistics payments.
    3. Review the "Apex Consulting" contract to see if the $4,999 payment is a recurring fixed fee.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Financial datasets often hide errors, policy breaches, or misconduct behind thousands of rows. This skill provides a systematic framework for detecting, triaging, and documenting financial anomalies with audit-ready rigor.

    What it does

    • Multi-method detection identifies duplicates, split transactions, timing outliers, and velocity spikes using statistical and rule-based tests.
    • Data profiling validates row integrity, detects referential gaps, and flags schema inconsistencies before analysis begins.
    • Risk-based triage prioritizes exceptions by financial exposure, recurrence, and evidence strength.
    • Evidence-based investigation reconstructs transaction timelines and traces related entries across ledgers and sub-ledgers.
    • Audit-ready reporting produces prioritized case registers, investigation memos, and control-monitoring specifications.

    How it works

    1. Intake and profiling: Define the population and reporting period while validating data quality and sign conventions.
    2. Baseline construction: Establish comparison groups based on historical medians, peer groups, or policy thresholds.
    3. Detection and triage: Apply relevant tests from the library and rank findings by potential financial impact.
    4. Investigation and reporting: Trace evidence for high-priority cases and generate a detailed findings memo with remediation steps.

    Frameworks & tools

    This skill is designed for analysts working with general ledgers, accounts payable/receivable, payroll systems, and bank reconciliations. It applies robust statistical methods like Median Absolute Deviation (MAD) and Interquartile Range (IQR).

    Why this beats prompting it yourself

    Generic prompts often lead to "hallucinated" fraud accusations or missed signals due to lack of financial context. This skill enforces strict evidence-first principles, separates confirmed from potential exposure, and follows a structured 8-phase investigation workflow used by professional auditors.

    Use cases

    • Reviewing journal entries for unusual manual postings or period-end adjustments.
    • Detecting split-purchase orders designed to bypass procurement approval limits.
    • Investigating duplicate vendor payments or sensitive master-data changes.
    • Reconciling bank statements against ledgers to find stale items or FX mismatches.

    Known limitations

    Does not perform autonomous financial transactions or filings. Requires manual data input or file access. Results are signals for human review, not definitive legal or disciplinary proof.

    Requires structured financial data. Cannot autonomously access private banking systems. Output is for investigation, not legal proof of fraud. Masking required for sensitive PII.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 8 days ago

    Permissions

    Write Files
    Listed8 days ago

    What's inside

    Frequently Asked Questions

    This skill provides a structured investigative framework that uses statistical methods like MAD and IQR to separate true anomalies from noise, ensuring every flag includes a clear audit trail rather than just a list of suspicious rows.

    The skill works with standard financial data exports (CSV, JSON, or Excel) from major ERP and accounting systems like SAP, Oracle, NetSuite, and QuickBooks, provided the AI agent has file-reading capabilities.

    You receive the full investigation workflow logic, the library of detection tests (including split-purchase and duplicate detection), and standardized templates for generating audit-ready investigation memos.

    No, the skill is designed to flag signals and reconstruct timelines for human review; it lacks the authority to execute payments or modify your live ledger, maintaining a secure 'read-only' analytical layer.

    Setup is immediate; once added to your AI agent, you simply provide your dataset and define your reporting period to begin the intake and profiling phase.

    New in Research & Analysis