GGUF Scanner Bypass PoC β Security Research Artifacts
WARNING: These are intentionally malformed/malicious model files for authorized security research. DO NOT load them outside sandboxed environments.
Purpose
Integer overflow, string overflow, negative dimensions, and path traversal in GGUF headers bypass modelscan 0.8.8 (6/6 MISSED). These target C parsers (llama.cpp, ggml) and can cause heap corruption, OOB reads, or DoS.
Responsible disclosure artifacts for the huntr MFV program.
Scanner Results (2026-03-20)
modelscan 0.8.8: ALL payloads MISSED.
| File | Technique | modelscan | Impact |
|---|---|---|---|
gguf_overflow_nkv.gguf |
n_kv set to 2^63-1 (max uint64) | MISSED | Heap allocation bomb β parser allocates massive array |
gguf_overflow_ntensors.gguf |
n_tensors set to 2^63-1 | MISSED | Heap allocation bomb |
gguf_string_overflow.gguf |
String length field set to 2^63-1 with short data | MISSED | OOB read / heap buffer overflow in string parsing |
gguf_tensor_traversal.gguf |
Path traversal ../../../etc/passwd in tensor name |
MISSED | File write if tensor name used as export path |
gguf_negative_dims.gguf |
Negative tensor dimensions (signed -1) | MISSED | Integer overflow in size calculation β heap corruption |
gguf_version_zero.gguf |
Invalid GGUF version=0 | MISSED | Parser undefined behavior on unsupported version |
Format
GGUF ($4,000 MFV bounty on huntr)
Researcher
vellaveto
- Downloads last month
- 93
Hardware compatibility
Log In to add your hardware
We're not able to determine the quantization variants.
Inference Providers NEW
This model isn't deployed by any Inference Provider. π Ask for provider support