CVE-2026-114114: Vyper through 0.5.0b1 Literal Value Substitution via Multi-byte Characters
Vyper 0.1.0b1 before 0.1.0b12, 0.1.0b14, and 0.4.1 through 0.5.0b1 mix character-based line offsets with Python's UTF-8 byte coloffset values in vyper/ast/parse.py, so multi-byte characters shift the source slices used to reconstruct later numeric literals. An attacker contributing source with non-ASCII identifiers that NFKC-normalize to ASCII, or an invisible byte-order mark, can silently substitute addresses, integers or other constants in compiled code.
Affected Software
Event History
Frequently Asked Questions
Who is realistically exposed to this issue?
Projects that compile Vyper source supplied or modified by an untrusted contributor are exposed, especially where review may not detect non-ASCII identifiers that normalize to ASCII or an invisible byte-order mark. The impact is on the compiled code: later numeric literals such as addresses or integers can be silently substituted.
What must an attacker do to exploit it?
The attacker needs to contribute Vyper source containing multi-byte characters before numeric literals. The described techniques are non-ASCII identifiers that NFKC-normalize to ASCII or an invisible byte-order mark, which shift source slicing during parsing and can alter reconstructed constants.
Which Vyper versions are affected?
Affected versions are 0.1.0b1 before 0.1.0b12, 0.1.0b14, and versions from 0.4.1 through 0.5.0b1. The provided information does not identify fixed versions for the 0.4.1 through 0.5.0b1 range.