CVE-2026-107717: medium-severity vulnerability in masci banks
Banks: User-controlled prompt input can be parsed as privileged chat messages
Published · Updated
E-mail alert when it lands in KEV, gets a public PoC or its score goes up.
13Vexday Risk Score
No sign of exploitation. No public exploitation artifact known so far.
ssvc Trackcvss 6.5epss 0.3%
Banks generates meaningful LLM prompts using a simple template language. Prior to 2.5.0, Banks Prompt.chat_messages() attempts to parse every line of rendered template output as ChatMessage JSON. When an application renders untrusted data and passes the returned ChatMessage objects to an LLM provider, attacker-controlled JSON can cross the prompt boundary and become a system, assistant, or tool message because ChatMessage.role accepts arbitrary strings. This can override application instructions, alter the intended prompt structure, or confuse downstream tool and message handling. This issue is fixed in version 2.5.0.
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N
Affected products
masci · banksRelated CVEs — masci banks
In the same product, most dangerous first.
CVE-2026-71492MEDIUMBanks: Path traversal in `DirectoryPromptRegistry.set()` allows arbitrary file write outside the registry rootEPSS 0.5%CVE-2026-44209HIGHBanks: Critical Remote Code Execution (RCE) via Jinja2 SSTIEPSS 0.4%CVE-2026-62663HIGHBanks: Arbitrary File Read via Path Traversal in Media Filters (image/audio/video/document)EPSS 0.4%CVE-2026-61536HIGHBanks: Unsafe importlib.import_module of attacker-controlled Tool.import_path in CompletionExtension allows RCEEPSS 0.4%CVE-2026-107716HIGHBanks: Symlink traversal and arbitrary file disclosure/overwrite in DirectoryPromptRegistryEPSS 0.1%