LayerFault/mlx-custom-code-suite

1---2license: apache-2.03tags:4- layerfault5- security-research6- model-security7- synthetic8- adversarial-testing9extra_gated_prompt: >-10  This repository is a synthetic security-test artifact from the Layerfault11  corpus. It intentionally contains adversarial characteristics (e.g.12  suspicious pickle opcodes, executable-format smuggling, prompt-injection13  strings) designed to exercise security scanner detection rules. It is14  **not** a usable ML model and must never be loaded or executed outside an15  isolated scanner-testing environment. By accepting, you confirm you16  understand this repository is a test fixture, not production model17  weights.18extra_gated_button_content: I understand this is a security test fixture and accept the risk19gated: auto20---21 22# mlx-custom-code-suite23 24> **SECURITY TEST ARTIFACT: DO NOT USE AS A PRODUCTION MODEL**25 26This repository is part of the Layerfault synthetic security corpus.27It is deliberately constructed to contain security-relevant characteristics for scanner testing.28 29**Corpus ID:** `LF-CORPUS-MLX-0001`30 31## Purpose32 33Minimal MLX-like package with custom Python code and deliberately partial profile metadata.34 35## Direct expected Layerfault rules36 37- None; this repository is a control/comparison input.38 39## Candidate rules40 41These are deliberately plausible targets that remain marked as candidates until the exact42Layerfault build used for certification confirms them.43 44- `LF-MLX-CUSTOM-CODE`45- `LF-MLX-PROFILE-INCOMPLETE`46 47## Negative-control rules48 49These should remain silent for this corpus item.50 51- None52 53## Safety54 55The corpus uses fake secrets, loopback/`.invalid` network destinations, harmless marker output,56and synthetic model behavior only. It is intended for static scanning and isolated security testing.57