VNX-343: Predictable Value Range from Previous Values
Detects secrets that are the previous value plus a small random step (token += random.randint(...), session_id = last + rng.nextInt(500), isn = last_isn + (rand() % 64000)). Python, Java, Kotlin, JavaScript/TypeScript, PHP, Ruby, Go, C and C++.
| Field | Value |
|---|---|
| Rule ID | VNX-343 |
| Severity | Medium |
| SARIF level | warning |
| Kind | sast |
| Languages | Python, Java, Kotlin, JavaScript / TypeScript, PHP, Ruby, Go, C, C++ |
| CWE | CWE-343 |
| CAPEC | CAPEC-66 |
| ATT&CK | T1190 |
| CWSS vector | CWSS:1.0/TI:M/AP:RU/AL:S/IC:N/FC:LT/RP:RU/RL:S/AV:L/AS:N/IN:A/SC:A/BI:M/DI:H/EX:M/EC:N/P:C |
| Tags | randomness predictable predictable-value-range cwe-343 |
| Source | internal/sast/rules/vnx-343.rego |
Overview
VNX-343 maps to CWE-343: Predictable Value Range from Previous Values. A value that moves forward by a small random step from the last one is predictable within a window: anyone who has seen one value knows the next lies just above it, and tries the few hundred candidates. This is the classic weak TCP initial-sequence-number design, and it appears in hand-rolled token and session-id generators.
The rule reports:
x += <random draw>wherexis named for a secret or an initial sequence number (token, session id, reset code, OTP, CSRF, password, API key, isn), and<previous> + <random draw>where the previous value is named last/prev/previous/current/old/next or for a secret, and the sum is stored under, returned as, or flows within six lines into a secret.
A random draw is random.randint/randrange/getrandbits/uniform, secrets.randbelow, Math.random() (also inside Math.floor), rand()/mt_rand()/random_int(), nextInt/nextLong (Random, ThreadLocalRandom, kotlin.random.Random), Go rand.Intn/IntN/Int63n/Int31n (also inside a numeric conversion), Ruby rand/random_number, and C rand()/random()/arc4random_uniform. A small step from a CSPRNG is still reported: the range is the weakness, not the generator.
What Gets Flagged
# FLAGGED
self.last_token += random.randint(1, 1000)
reset_code = prev_code + random.randrange(10, 99)
session_id = last_session + secrets.randbelow(256)
// FLAGGED
long sessionId = previousSessionId + ThreadLocalRandom.current().nextInt(1, 500);
/* FLAGGED: initial sequence numbers a small step apart */
isn = last_isn + (rand() % 64000);
tcb->snd_isn += rand() & 0xffff;
What Is Not Flagged
- Jitter, delays, backoff, poll intervals, positions, totals and other values that are not secrets (
self.retry_delay += random.uniform(0, 1),next_poll = last_poll + random.randint(1, 5)). - Counts and bookkeeping names (
token_count,session_timeout). - A draw written before the previous value (
rand() + last), and an increment by a constant (VNX-342). - Test files, and the pattern in comments or string literals.
Remediation
- Generate each secret independently with a CSPRNG over its full width:
secrets.token_urlsafe(),crypto.randomBytes(),random_bytes(),SecureRandom,crypto/rand. - For sequence numbers, use a keyed hash of the connection identifiers plus a clock (RFC 6528) rather than a random step from the previous value.
- For confirmed false positives, add a
# vulnetix-ignore: VNX-343comment on the line.
References
- CWE-343: Predictable Value Range from Previous Values
- RFC 6528: Defending against Sequence Number Attacks
Suppressing this rule
Put a nosec comment on the line, or on line 1 to cover the whole file. Use the comment marker of the file’s language (#, //, -- or ;):
# nosec vnx-343
// nosec vnx-343
Or record an ignore rule: vulnetix ignore add --rule vnx-343 --reason "...". See Suppressing findings.