LF Security Corpus: Hard Negative Controls
Near-miss fixtures for false-positive resistance and detector precision. Synthetic security fixtures, not usable models.
UpdatedNote LF-CH-NEG-0006: Binary contains ZIP local-header magic but is not a valid archive.
LayerFault/near-miss-credential-placeholder
UpdatedNote LF-CH-NEG-0008: Synthetic placeholder resembles a credential-shaped value but is explicitly fake.
LayerFault/near-miss-js-child-process-doc
UpdatedNote LF-CH-NEG-0003: JavaScript comment documents child_process without using it.
LayerFault/near-miss-pickle-magic-text
UpdatedNote LF-CH-NEG-0005: Text includes escaped Pickle-like bytes and dangerous-global terminology.
LayerFault/near-miss-python-eval-string
UpdatedNote LF-CH-NEG-0002: Python string documents eval syntax without executing it.
LayerFault/near-miss-python-subprocess-comment
UpdatedNote LF-CH-NEG-0001: Python comments mention subprocess APIs but contain no executable call site.
LayerFault/near-miss-shell-curl-doc
UpdatedNote LF-CH-NEG-0004: Text documentation contains curl-pipe-shell syntax.
LayerFault/near-miss-template-security-words
UpdatedNote LF-CH-NEG-0007: Text mentions template introspection primitives without being a template.