Security review program
Every site and app we ship goes through the same checklist before it carries the seal — the same rigor we apply to Snipe itself. If it wears the roundel, it passed. This page is the rubric, the grade bands, and the latest report card.
Six categories, each scored 0–100 against checks we can actually verify — response headers, cookie flags, rate-limit behavior, code review of every place user data touches the page, and a live round-trip through the auth flow on a throwaway database. The overall score is the weighted average.
Meets the bar on every category. Earns the seal.
Ships with noted gaps and a fix-by date. No seal until resolved.
Does not ship. Findings go back to the build queue.
HTTPS everywhere, HSTS with includeSubDomains, a real Content-Security-Policy (no blanket unsafe-inline on scripts), frame-ancestors/X-Frame-Options, nosniff, and a tight referrer policy.
Passwords hashed with scrypt and a per-user salt. Session cookies are HttpOnly, SameSite, and Secure on HTTPS, scoped to API paths only, with the server storing a hash of the token — never the token itself. Sign-in rotates the session; sign-out kills it. New accounts prove they own their email before activating.
Per-IP rate limits on every public endpoint, tighter ones on sign-in, a capped number of wrong verification codes before reset, honeypot fields on open forms, and input length caps so junk can't balloon the database.
Code review of every place data reaches the page — textContent over innerHTML for anything dynamic, parameterized SQL only, no eval, no inline event handlers. Cross-origin calls carry an Origin check on top of SameSite cookies.
No credentials or tokens in localStorage, minimum-viable personal data collected, nothing sensitive written to logs, no third-party trackers, and secrets live as server-side environment variables — never in shipped code.
The site doesn't overclaim — approximate data is labeled approximate, assessor parcel lines aren't presented as legal surveys, demo content isn't passed off as real, and this methodology is public.
| Category | What we verified | Result |
|---|---|---|
| Transport & headers20% | HSTS max-age=15552000; includeSubDomains · CSP denies inline script handlers, object-src 'none', form-action 'self', frame-ancestors 'self' · nosniff · referrer-policy: no-referrer |
PASS |
| Authentication & sessions25% | scrypt + per-user salt · verification code SHA-256-hashed, 10-min expiry, 8-attempt cap · session = 256-bit random token, SHA-256 stored · cookie HttpOnly; SameSite=Lax; Secure; Path=/api · no enumeration on duplicate signup |
PASS |
| Abuse controls15% | Sign-in locked after limit hit in live test (10/15 min) · register/verify 15/hr · leads 10/hr · global API 300/15 min · honeypot fields on every open form · inputs length-capped server-side | PASS |
| Injection surface15% | Dynamic DOM writes use textContent · all queries parameterized via node:sqlite · no eval, no inline handlers · Origin check on mutating account routes |
PASS |
| Data hygiene15% | Accounts store name + email + saved listings only · nothing in localStorage but display prefs · secrets via Fly secrets · no analytics/tracker scripts | PASS |
| Honesty & disclosure10% | County plat lines labeled as assessor data, not survey-grade · demo listings marked as such pending MLS data · private-preview pages carry noindex,nofollow |
PASS |