Revue de sécurité des systèmes financiers backend

---

name: security-review-backend description: Analyze the backend architecture and data persistence layer to identify security vulnerabilities in the handling of financial information.

Stack Context

• Language / Framework: C# / ASP.NET Core Web API
• ORM: Entity Framework Core with global query filters for ownership isolation
• Authentication: Auth0 JWT — user_id / sub claim is the identity anchor
• Infrastructure: PostgreSQL, Redis, Docker

Where to Look

Focus your search on these file patterns:

• *Controller.cs — authorization enforcement, model binding, input validation
• *Handler.cs / *Service.cs — business logic, data access, ownership filters
• *Configuration.cs / *EntityTypeConfiguration.cs — EF column mappings (field types, encryption)
• *DbContext.cs — global query filters, ownership constraints
• appsettings*.json, Program.cs — middleware pipeline, logging sinks, secret handling

Analysis Guidelines

Encryption at Rest

• Identify numeric or text fields representing monetary values (e.g. balance, amount, salary) and verify they are not stored in plain text in the database.
• Detect missing symmetric encryption (e.g. AES-256) for sensitive financial columns in EF configurations.

Identity Isolation (Auth0 Integration)

• Verify that the user_id / sub from the Auth0 token is the sole link to financial records.
• Ensure no tables directly associate personally identifiable data with balances without an abstraction layer.
• Check that global EF query filters enforce ownership and cannot be bypassed (e.g. via IgnoreQueryFilters()).

Object-Level Authorization (BOLA)

• Audit controllers and handlers to confirm every query filters strictly by the authenticated user's ID.
• Flag any endpoint that accepts a user or resource ID from the request body or URL without re-validating ownership against the token claim.

Mass Assignment

• Detect model binding that accepts raw domain entities or overly permissive DTOs, allowing callers to set fields like UserId, Balance, or Role directly.

Log and Exception Leakage

• Verify that error-handling and logging routines do not capture or serialize objects containing financial amounts to server logs.
• Check that exception middleware does not return stack traces or internal model data to API consumers.

Rate Limiting

• Flag financial write endpoints (transfers, payments, balance updates) that lack rate limiting or idempotency controls.

Integrity Validation

• Detect missing business-logic validations (e.g. allowing negative amounts where not permitted, or numeric overflow in financial calculations).
• Check that decimal precision is enforced at the domain level, not only at the database column level.

Output Instructions

For each risk found, report:

| Field | Detail | |---|---| | File | Path to the affected file | | Line | Line number(s) | | Description | Technical description of the vulnerability | | Priority | Critical / High / Medium | | Recommendation | Suggested fix or mitigation |

Report File

Save the report as a Markdown file under:

docs/security-audits/security-audit-{scope}-YYYY-MM-DD.md

• Use today's date in the filename.
• {scope} identifies what was reviewed. Use backend for a general review or a more specific segment when the review is scoped (e.g. backend-authentication, backend-persistence, backend-application-commands, backend-application-queries, backend-domain).
• If a file for today with the same scope already exists, append a counter (e.g. -2).