X-Ray Any Binary File Locally.
Zero Upload. Instant Schema.
Drop a 70 GB .gguf model, a 20 GB .parquet dataset, a sensitive .sqlite / .har dump, or a corrupted .xlsx download. Our paged header/footer reader decodes internal schemas & magic bytes in <20ms without uploading a single byte.
Drop any local file (1 KB to 100 GB+) to inspect binary headers & schema
Reads tiny File.slice() header/footer pages in a sandboxed Web Worker · 0 Bytes Uploaded
Instant 100 GB+ File Inspection
Unlike naive hex viewers that crash browser RAM or cloud tools that demand uploads, our adaptive cursor slices only dynamic header & footer pages (GGUF KV tables, SafeTensors JSON, Parquet Thrift footers, ZIP EOCD).
Catch Spoofed & Broken Files
Distinguishes .xlsx, .ipa, .aab, .npz, .usdz and PyTorch .pt inside shared ZIP headers—and flags API gateway HTML errors saved with binary extensions.
Edge-Case Samples + SHA-256
Grab 1 KB unit-test stubs, UTF-8 BOM multiline CSVs, EXIF Orientation=6 JPEGs, multi-column RAG PDFs, and truncated footers via immutable curl links.
Explore Modern High-Value File Specifications
Inspect 70GB+ tensor weights, quantization KV metadata & pickle safety locally
Decode Parquet Thrift footers, SQLite schemas, Arrow IPC & Avro headers offline
Verify ISOBMFF brands, EXIF rotation traps, AI PNG prompts & glTF/USDZ scenes
Audit WASM sections, sanitize HAR captures, dissect mobile bundles & detect spoofed files
QA Upload Security & AI RAG Parser Torture Suite
Benchmarking Docling, Marker, LlamaParse, or hardening your SaaS file upload gateway? Get 500+ curated edge-case files—multi-column cross-page financial PDFs with 100% human-verified Ground-Truth Markdown/JSON outputs, benign polyglot headers, encoding edge-cases, and automated pytest / vitest runners.