Only the latest release receives fixes.
Please do not open a public issue. Use GitHub's private vulnerability reporting instead: Security → Report a vulnerability on this repository. You can expect an acknowledgement within 5 working days; if the report is confirmed, a fix or mitigation plan follows within 30 days.
Please include the affected version or commit, steps to reproduce and the impact you observed.
See docs/threat-model.md for assets, threats and mitigations. In short:
| Area | Default |
|---|---|
| Price history | Append-only at the database level: the API role has no UPDATE, DELETE or TRUNCATE on price_observation; observed_at only moves forward in current_price |
| Database access | The API runs as raf_app (DML only, also on Hangfire's schema; no DDL); migrations and the demo seed run in a one-shot process as raf_migrator |
| Feed input | Per-item plausibility checks, pack-size cross-check, XML with DTDs prohibited (no XXE), parameterised SQL only |
| Personal data | None processed: baskets are anonymous, identified by a device-generated UUID |
| Input limits | At most 100 basket lines, quantities 1–99, 200 watched products per SignalR connection |
| Containers | Chiseled .NET runtime images and unprivileged nginx, no shell or package manager in the .NET images, non-root |
| Secrets | None in the repository; compose defaults are for local development and can be overridden via .env |
| Supply chain | Centrally pinned package versions (NuGet and the app's package-lock.json), updated by hand; CI fails on known-vulnerable packages (dotnet list package --vulnerable --include-transitive, npm audit --omit=dev --audit-level=high); CodeQL security-and-quality on every push |
Not included: authentication and authorisation, rate limiting, TLS termination. The Hangfire dashboard and the simulator's admin endpoints are for the local demo only. This repository is a demonstration with synthetic data and must not be exposed as-is.