NLTK Path Security Bypass Enables Arbitrary Local File Read

NLTK before 3.10.0 contains an arbitrary local file read vulnerability in StreamBackedCorpusView that bypasses pathsec.ENFORCE by calling builtins.open() directly instead of pathsec.open(). Attackers who control the fileid argument can read arbitrary local files regardless of the ENFORCE setting, including sensitive system files and application credentials.

Casky was already ahead

This CVE exploits attack patterns that Casky's 351matched skills already investigate — long before this vulnerability was disclosed. Claude's reasoning model maps these techniques to MITRE ATT&CK, so practitioners who ran these skills have already seen the threat behaviour in their findings.

Analysis

CVE-2026-63312 is a critical path traversal vulnerability in the Natural Language Toolkit (NLTK) library that undermines its built-in security controls. StreamBackedCorpusView circumvents pathsec.ENFORCE protections by calling Python's native builtins.open() directly rather than routing through the secure pathsec.open() wrapper. This allows attackers controlling the fileid parameter to read any file on the system—including sensitive configurations, private keys, and application credentials—completely bypassing intended security boundaries. Any application using NLTK versions before 3.10.0 that processes untrusted fileid inputs is vulnerable, making this especially critical for NLP applications, chatbots, and text processing services that may accept user-supplied corpus references.

Casky's 351 mapped skills detect the attack patterns underlying this vulnerability across two MITRE ATT&CK techniques: TA0001 (Initial Access) and TA0007 (Discovery). The extended reasoning capabilities in Claude AI identify the behavioral signatures—such as unexpected file access patterns, direct syscall usage bypassing security wrappers, and credential material appearing in application logs or memory. Practitioners using Casky would see findings highlighting: (1) instances where untrusted input flows directly to file operations without validation, (2) calls to builtins functions that circumvent security middleware, and (3) access attempts to sensitive files outside intended corpus directories. The skills framework connects these observations to the underlying CWE-22 path traversal pattern, enabling teams to correlate suspicious file access with potential CVE-2026-63312 exploitation attempts before credentials are exfiltrated.

Live Threat IntelligenceComing soon

Casky Risk Score

—

EPSS Probability

—

Shodan Exposure

—

GreyNoise Signal

—

Composite risk scoring from EPSS, CISA KEV, Shodan, and GreyNoise — 21 security APIs correlated into a single Casky Risk Score. Coming in Casky Pro. Join early access →

351 Casky skills cover this attack pattern

These skills use Claude AI's reasoning model to surface findings in the same attack categories as CVE-2026-63312.

Abuse of Public-Facing API: Mobile API Abuse

mobile security · low

Run

abusing-dpapi-for-credential-access

red teaming · high

Run

abusing-shadow-credentials-for-privesc

red teaming · high

Run

MITRE ATT&CK Techniques

TA0001 TA0007

Investigate the attack patterns behind CVE-2026-63312

Casky has 351 skills that investigate the attack patterns behind CVE-2026-63312. Run one and get CVSS-scored findings in 3 minutes.

Run the skill that detects this →

X Instagram LinkedIn