Back to database
Soon65Medium6.5VulnerabilityCVE-2026-107717No patch link observed

CVE-2026-107717 — Banks generates meaningful LLM prompts using a simple template language.

Published Oct 8, 2026, 10:17 PM UTCIngested 38m agoSource NVD(cve-db)CVE-2026-107717

Description

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 v3.1 base metrics

CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N
6.5

Medium severity

Band computed from the CVSS base score, not the source's own label — so it means the same thing across every feed.

AV

Network

Attack Vector

AC

Low

Attack Complexity

PR

None

Privileges Required

UI

None

User Interaction

S

Unchanged

Scope

C

Low

Confidentiality

I

Low

Integrity

A

None

Availability