Highflame ramparts EUVD-2026-114723

Exploitation requires the attacker to distribute a skill bundle containing an unchecked-hash .pyc planted specifically under scripts/<subdir>/__pycache__/, and the victim to (a) scan that bundle with a vulnerable ramparts version in the 0.8.1-0.8.10 range, whose collect_bundle_siblings and walk_bundle_subdir routines silently ignore nested directories and .pyc files, and (b) actually run the skill so CPython loads and executes the hidden bytecode. Limiting factors: passive user interaction is required (the victim must obtain and run the attacker-supplied bundle), only ramparts 0.8.1 through 0.8.10 are affected, and the payload must be placed in the exact nested __pycache__ path that the scanner skips; there is no path to triggering this without victim-side adoption of the malicious bundle.

The input CVSS 4.0 base score is 7.1 (High) with vector AV:N/AC:L/AT:N/PR:N/UI:P - network-distributable, low complexity, no privileges, but requiring passive user interaction (the victim must adopt and run an attacker-supplied skill bundle). No CISA KEV listing is present, so this is not confirmed actively exploited; no EPSS score was provided, so exploitation probability is unquantified; and no verified working exploit code was found (references are a VulnCheck advisory, GitHub issue #154, and a source-code line). The vuln.today decision label is 'this_week', indicating prompt but non-emergency remediation - consistent with the absence of KEV/active-exploitation signals. The real-world risk profile is that of a supply-chain / protection-mechanism bypass (CWE-693): it is genuinely concerning because it silently defeats a tool teams rely on to vet skill bundles, but exploitation is targeted rather than wormable because it depends on a victim running a malicious bundle through a vulnerable ramparts version (0.8.1-0.8.10). One signal conflict worth noting: the supplied 4.0 vector scopes impact to integrity only (VI:H, VC:N, VA:N), yet the described mechanism is arbitrary CPython execution of an unchecked .pyc, which realistically produces fuller code-execution impact than an integrity-only scoring implies - verify impact scope with the vendor advisory.

An attacker publishes or shares a skill bundle that hides a malicious Python bytecode file at scripts/<subdir>/__pycache__/. A victim scans the bundle with ramparts 0.8.1-0.8.10, which reports it clean because collect_bundle_siblings and walk_bundle_subdir skip nested directories and .pyc files, and then runs the skill, at which point CPython loads and executes the hidden payload. No verified public exploit code was identified at time of analysis; references point to a VulnCheck advisory and GitHub issue #154 rather than a confirmed working exploit.