← All projects
SOFTWARE + GOVERNANCE, RISK & COMPLIANCE

Community Bank CRM
& Risk Assessment

One project, two perspectives: build a banking workflow, then assess what it would take to protect real customer data. Explore 25 fictional clients across Indiana, leads, meetings, location-aware lending offers, and live fraud-note scoring, alongside a source-based review of eleven risk scenarios, proposed controls, and two findings that were fixed and re-tested.

  • Fictional clients and data
  • Illustrative risk ratings
  • Most proposed controls are not implemented
JavaScriptBanking workflowsRisk assessmentControl evidenceTreatment planning

Sample-data prototype with simulated roles and browser-local notes. Add a Fraud Note to recalculate its score. The linked GRC exercise assesses a hypothetical production deployment; proposed controls are not implemented.

BUILD · ASSESS · IMPROVE
  1. 01 / EXPLORETry the banking workflows

    Client lookup · leads · meetings · fraud scoring

  2. 02 / ASSESSTrace risks to the source

    Eleven scenarios linking code evidence to business impact

  3. 03 / PLANDocument the path forward

    Proposed controls, treatment updates, and assessment export

Implementation notes: architecture behind the prototype

TECHNICAL DOCUMENTATION / VERSION CONTEXT

Architecture behind the prototype

My June 2026 technical documentation describes a fuller Node.js and SQLite prototype. The public portfolio runs a static adaptation so visitors can explore sample workflows without a server account. These versions have different security boundaries.

  1. DOCUMENTED FRONTEND

    Browser interface

    HTML, CSS, JavaScript, and Fetch-based calls for customer, lead, meeting, and note workflows.

  2. DOCUMENTED BACKEND

    Application API

    The document describes password hashing, expiring bearer sessions, and permission checks on API operations.

  3. DOCUMENTED DATABASE

    SQLite records

    Related customer and workflow tables, a role-permissions table, and paginated queries.

I documented the backend architecture; the public demo simulates roles and does not run those backend controls. My next step is to validate access enforcement with source review and API tests, then complete edit operations and audit logging.

FOUND, FIXED, AND RE-TESTED

Audit the output, then fix it.

A follow-up audit planted markup in every browser-saved field and loaded every page. It found stored markup injection on the Leads, Meetings, and Pipeline pages (R11). I escaped every record field written into an HTML template, encoded identifiers in URLs, and added a test that fails if a field is written unescaped. Bank notes (R03) were re-verified. Both now show a residual rating in the assessment.

A review of Discover Needs fixed two more defects. The client page counted a hard-coded list (3) while the Discover page counted the engine's output (13); both now read one recommendation list. A home-equity offer appeared for a client whose profile showed $0 equity; it now requires estimated equity of at least $50,000 and a loan-to-value of 85% or less, and products that were held back are listed with the reason.

These fixes address specific findings. They do not establish that the application is secure or close the other risks, and the assessment remains pinned to its historical source snapshot.

Next verification priorities for the documented backend are denied-request tests for each role, record-level access checks, session expiry and logout tests, and evidence of security event logging.