Pylint EUVD-2026-108257

Exploitation requires PYLINT_HOME to resolve to a shared, group-writable, or otherwise untrusted directory AND the attacker to have write access to that cache directory so they can overwrite the predictable module-statistics pickle file; the victim must then run Pylint (any version from 0.15.2 up to but not including 4.0.9), which reloads the cache via pickle.load. The exploit is explicitly NOT possible against the default configuration: the default cache ~/.cache/pylint is writable only by the user running Pylint, so this is a multi-user/shared-host and CI-runner concern rather than a default-install flaw. Limiting factors: local access only, attacker needs a foothold on the host, a non-default or misconfigured shared PYLINT_HOME, and victim-triggered execution (the next Pylint run).

This is a genuine CWE-502 pickle-deserialization code-execution flaw (CVSS 4.0 base 5.4, Medium) that is correctly rated as a real but conditional issue rather than an emergency - the vuln.today Decision label is 'this_month', not act_now/emergency. The supplied CVSS 4.0 vector tells a consistent story: local vector (AV:L), high attack complexity (AC:H), a present attack requirement (AT:P), low privileges (PR:L), and passive user interaction (UI:P) gated behind high confidentiality/integrity/availability impact (VC:H/VI:H/VA:H) with no scope/subsequent-system effect. No EPSS score and no KEV status were provided; absence of KEV means this is NOT confirmed as actively exploited, and the lack of an EPSS figure means widespread-exploitation probability cannot be quantified from the input - state this as a gap rather than assuming low risk. The references are the GitHub Security Advisory, two fix commits, and the release tag; none is an exploit, and the input does not mark any PoC as 'Publicly available (confirmed)', so there is no public exploit identified at time of analysis. The decisive risk limiter is architectural: exploitation requires an attacker-writable PYLINT_HOME, and the default per-user cache (~/.cache/pylint) is writable only by the running user, so the attack surface collapses to multi-user hosts with a shared or group-writable cache. Net: a legitimately impactful RCE in a hugely popular tool, but a narrow precondition keeps real-world priority moderate - patch on a normal cycle, prioritize shared-host/CI environments.

On a multi-user build server where PYLINT_HOME points to a shared, group-writable cache directory, a low-privileged attacker overwrites the predictable stats cache file with a crafted pickle payload. When another user (e.g. a developer or a CI service account) next runs Pylint, pickle.load deserializes the malicious file and executes arbitrary code with that victim's privileges. No public exploit is identified in the input, but the pickle-RCE technique is well understood and the fix commit confirms the vulnerable code path.