P02 Lab 11 — Password Security & Breach Check

WARMUP: Chapter 15 (Cryptographic Failures); NIST 800-63.

The idea

Two engineering controls around passwords: store them with a slow, salted, memory-hard KDF (Argon2id/bcrypt/scrypt/PBKDF2 — never a fast hash like SHA-256), and screen them against breach corpora without leaking them. The breach check reproduces Have I Been Pwned's k-anonymity range query: the client sends only the first 5 hex chars of the SHA-1 hash (a prefix shared by thousands of breached hashes), the server returns the matching suffixes, and the client matches locally — so the password and its full hash never leave the client. It's a lovely intersection of crypto and privacy (the Privacy module's k-anonymity, applied to a real API).

What you build (lab.py)

  • kdf_strength(algo, iterations) — unacceptable (fast/unsalted hashes) / weak (PBKDF2 below the iteration floor) / strong (Argon2id/bcrypt/scrypt, or PBKDF2 with enough iterations).
  • password_strength(pw) — deterministic length × character-class rating.
  • sha1_hex / hash_prefix — the HIBP hash format and the 5-char prefix split.
  • range_query(prefix, db) — server side: returns {suffix: count} for a prefix only.
  • check_breach(pw, db) — client side: uses only a range query; returns the seen count (0 = safe).

Cases the tests cover

  • KDF classification (SHA-256 unacceptable for passwords; PBKDF2 below/above the iteration floor; Argon2id strong); strength bands; the range query returns only same-prefix suffixes; breach found vs missed; and explicitly that the check works from the prefix alone (privacy-preserving).

Run

pip install -r requirements.txt
LAB_MODULE=solution pytest -q
pytest -q

Hardening / extensions

  • Estimate cracking cost/time from the KDF and parameters; add per-algorithm tuning guidance.
  • Use a real Argon2/bcrypt library and verify salt uniqueness and rehash-on-login-after-upgrade.
  • Add passphrase/zxcvbn-style strength and a denylist of breached/common passwords at signup.

Interview / resume

"Built password-storage KDF classification and a privacy-preserving breach check that mirrors Have I Been Pwned's k-anonymity range query — the password and its full hash never leave the client — tying crypto and privacy to a real-world API."

Limitations: strength rating is a simple heuristic (not zxcvbn); breach check uses SHA-1 (the HIBP corpus format) as a lookup key only, not a security control.