# fileformat.co

> Privacy-First Local File Inspector, Compiler-Verified Binary Spec Atlas & Deterministic QA Sample Registry.

## Core Machine-Readable Endpoints
- Magic Bytes Global Dictionary (JSON): https://fileformat.co/api/v1/magic-bytes.json
- Full QA Sample Fixtures Index (JSON): https://fileformat.co/api/v1/samples/index.json
- Per-Format Specification API Pattern: https://fileformat.co/api/v1/ext/{slug}.json
- Full Inline Specification Dump: https://fileformat.co/llms-full.txt
- Local Zero-Upload Browser Inspector: https://fileformat.co/inspect
- Local Zero-Upload HAR Sanitizer: https://fileformat.co/tools/har-sanitizer

## Deterministic Sample URL & Checksum Catalog
All sample URLs under `https://fileformat.co/fixtures/v1/` are CORS-enabled, immutable (v1 is frozen forever), and CC0 Public Domain for CI/CD & unit tests.

### AI & Model Weights
- [.gguf (GGUF)](https://fileformat.co/ext/gguf): Self-contained binary model weight format for llama.cpp and local LLM inference engines, storing architecture KV metadata, BPE vocabulary, and quantized tensor blocks in a single mmap-ready file. | Magic: `47 47 55 46` | Spec JSON: https://fileformat.co/api/v1/ext/gguf.json | Minimal Fixture (1120 B, sha256:2ec356e162053c78...): https://fileformat.co/fixtures/v1/gguf/minimal-llama.gguf
- [.jsonl (JSONL)](https://fileformat.co/ext/jsonl): Streaming line-oriented text format where every newline-separated record is a standalone UTF-8 JSON value—standard for LLM fine-tuning (SFT/DPO) datasets and batch APIs. | Magic: `structural` | Spec JSON: https://fileformat.co/api/v1/ext/jsonl.json | Minimal Fixture (308 B, sha256:85b7b589f6426c99...): https://fileformat.co/fixtures/v1/jsonl/minimal-sft.jsonl
- [.npy (NPY)](https://fileformat.co/ext/npy): Simple, mmap-friendly binary representation of a single homogeneous dense tensor with an explicit ASCII header dictionary describing dtype, byte order, and shape. | Magic: `93 4E 55 4D 50 59` | Spec JSON: https://fileformat.co/api/v1/ext/npy.json | Minimal Fixture (256 B, sha256:fd4257fe798868b2...): https://fileformat.co/fixtures/v1/npy/minimal-f32-4x8.npy
- [.npz (NPZ)](https://fileformat.co/ext/npz): ZIP container bundling multiple named NumPy `.npy` tensor arrays, widely used for embeddings, calibration datasets, and JAX/MLX checkpoints. | Magic: `50 4B 03 04` | Spec JSON: https://fileformat.co/api/v1/ext/npz.json | Minimal Fixture (668 B, sha256:5df985f2575c4c4a...): https://fileformat.co/fixtures/v1/npz/minimal-weights.npz
- [.pt (PyTorch PT)](https://fileformat.co/ext/pt): Default PyTorch `torch.save()` archive wrapping a Python Pickle instruction stream (`data.pkl`) alongside raw tensor storage files inside a ZIP container. | Magic: `50 4B 03 04` | Spec JSON: https://fileformat.co/api/v1/ext/pt.json | Minimal Fixture (1829 B, sha256:64f990b860de211a...): https://fileformat.co/fixtures/v1/pt/minimal-checkpoint.pt
- [.safetensors (SafeTensors)](https://fileformat.co/ext/safetensors): Hugging Face's memory-safe, zero-copy tensor serialization format designed to replace Pickle-based PyTorch checkpoints without arbitrary code execution risks. | Magic: `7B 22` | Spec JSON: https://fileformat.co/api/v1/ext/safetensors.json | Minimal Fixture (432 B, sha256:0d9894b505b77ec0...): https://fileformat.co/fixtures/v1/safetensors/minimal-bf16.safetensors
- [.onnx (ONNX)](https://fileformat.co/ext/onnx): Cross-framework Protocol Buffers computation graph and weight format used by ONNX Runtime, TensorRT, and CoreML converters. | Magic: `08` | Spec JSON: https://fileformat.co/api/v1/ext/onnx.json | Minimal Fixture (319 B, sha256:aa4e82d69b4579c1...): https://fileformat.co/fixtures/v1/onnx/minimal-linear.onnx

### Data Engineering & Columnar
- [.arrow (Arrow IPC)](https://fileformat.co/ext/arrow): Zero-copy in-memory columnar format serialized to disk with 8-byte aligned `ARROW1\x00\x00` framing and FlatBuffers schema/record-batch descriptors. | Magic: `41 52 52 4F 57 31 00 00` | Spec JSON: https://fileformat.co/api/v1/ext/arrow.json | Minimal Fixture (666 B, sha256:b86279b7e38e81af...): https://fileformat.co/fixtures/v1/arrow/minimal-ipc.arrow
- [.avro (Avro)](https://fileformat.co/ext/avro): Row-oriented binary data serialization container widely used in Kafka/Confluent event streaming and Iceberg manifest lists, embedding full JSON writer schema in the file header. | Magic: `4F 62 6A 01` | Spec JSON: https://fileformat.co/api/v1/ext/avro.json | Minimal Fixture (198 B, sha256:8f5420609b6bb63e...): https://fileformat.co/fixtures/v1/avro/minimal-null.avro
- [.csv (CSV)](https://fileformat.co/ext/csv): Plain-text tabular interchange format notorious for real-world parser edge cases: optional UTF-8 BOMs (`EF BB BF`), multiline quoted cells, doubled quote escapes (`""`), and formula injection. | Magic: `structural` | Spec JSON: https://fileformat.co/api/v1/ext/csv.json | Minimal Fixture (89 B, sha256:05718c2daf0ae301...): https://fileformat.co/fixtures/v1/csv/minimal-rfc4180.csv
- [.feather (Feather v2)](https://fileformat.co/ext/feather): Fast, language-agnostic DataFrame persistence format for Python (Pandas/Polars) and R, standardized since v2 directly atop the Apache Arrow IPC file specification. | Magic: `41 52 52 4F 57 31 00 00` | Spec JSON: https://fileformat.co/api/v1/ext/feather.json | Minimal Fixture (698 B, sha256:a7b7e3734e10f277...): https://fileformat.co/fixtures/v1/feather/minimal-v2.feather
- [.parquet (Parquet)](https://fileformat.co/ext/parquet): De-facto standard hybrid columnar storage format for modern data lakes (Iceberg, Delta, DuckDB, Spark), placing Thrift Compact Protocol schema and statistics in the file trailer. | Magic: `50 41 52 31` | Spec JSON: https://fileformat.co/api/v1/ext/parquet.json | Minimal Fixture (1131 B, sha256:8b84a4eb04fdad84...): https://fileformat.co/fixtures/v1/parquet/minimal-snappy.parquet
- [.sqlite (SQLite 3)](https://fileformat.co/ext/sqlite): Ubiquitous single-file ACID relational B-tree database with a big-endian 100-byte header at offset 0x00 and full SQL DDL catalog stored in Root Page 1 (`sqlite_schema`). | Magic: `53 51 4C 69 74 65 20 66 6F 72 6D 61 74 20 33 00` | Spec JSON: https://fileformat.co/api/v1/ext/sqlite.json | Minimal Fixture (12288 B, sha256:dee45e02f67ee69f...): https://fileformat.co/fixtures/v1/sqlite/minimal-users.sqlite
- [.duckdb (DuckDB)](https://fileformat.co/ext/duckdb): Single-file vectorized OLAP database format storing compressed columnar row-groups in 4 KB / 256 KB block-managed pages. | Magic: `44 55 43 4B` | Spec JSON: https://fileformat.co/api/v1/ext/duckdb.json | Minimal Fixture (536576 B, sha256:465f054e7e501c94...): https://fileformat.co/fixtures/v1/duckdb/minimal-v64.duckdb
- [.orc (Apache ORC)](https://fileformat.co/ext/orc): Stripe-based columnar storage format originating in the Hive/Trino/Presto ecosystem with Protobuf-encoded Footer and a 1-byte PostScript length at the very last byte of the file. | Magic: `4F 52 43` | Spec JSON: https://fileformat.co/api/v1/ext/orc.json | Minimal Fixture (344 B, sha256:09713ef2158b8196...): https://fileformat.co/fixtures/v1/orc/minimal-stripe.orc

### Modern Media & Spatial 3D
- [.avif (AVIF)](https://fileformat.co/ext/avif): Royalty-free HDR & wide-gamut next-gen web image format wrapping AV1 keyframes inside an ISO Base Media (`ftyp`) container (`major_brand` = `avif` or `avis`). | Magic: `66 74 79 70 61 76 69` | Spec JSON: https://fileformat.co/api/v1/ext/avif.json | Minimal Fixture (2446 B, sha256:c1d62b9cfb1ac1ef...): https://fileformat.co/fixtures/v1/avif/minimal-128x128.avif
- [.glb (glTF GLB)](https://fileformat.co/ext/glb): Khronos Group's single-file binary 3D delivery format combining a 12-byte header, a UTF-8 JSON scene-graph chunk (`0x4E4F534A`), and a raw GPU geometry/texture buffer (`0x004E4942`). | Magic: `67 6C 54 46 02 00 00 00` | Spec JSON: https://fileformat.co/api/v1/ext/glb.json | Minimal Fixture (316 B, sha256:f9316dd27503e6ed...): https://fileformat.co/fixtures/v1/glb/minimal-triangle.glb
- [.heic (HEIC)](https://fileformat.co/ext/heic): Apple iOS default camera photo container encapsulating HEVC (H.265) intra-coded image tiles, depth maps, and EXIF metadata inside an ISO Base Media (`ftyp`) box hierarchy. | Magic: `66 74 79 70 68 65 69 63` | Spec JSON: https://fileformat.co/api/v1/ext/heic.json | Minimal Fixture (1192 B, sha256:a391bd763c271131...): https://fileformat.co/fixtures/v1/heic/minimal-64x64.heic
- [.jpg (JPEG)](https://fileformat.co/ext/jpg): Marker-segmented DCT lossy image format (`FF D8 FF ... FF D9`), where embedded APP1 EXIF TIFF headers carry camera rotation flags (`Orientation=6`) and GPS coordinates. | Magic: `FF D8 FF` | Spec JSON: https://fileformat.co/api/v1/ext/jpg.json | Minimal Fixture (644 B, sha256:72ad937c3cc13695...): https://fileformat.co/fixtures/v1/jpg/minimal-8x8-jfif.jpg
- [.png (PNG)](https://fileformat.co/ext/png): Lossless CRC-checked chunked raster image format, heavily used in AI workflows (ComfyUI / Stable Diffusion) to store generation graphs and prompts inside `tEXt`/`iTXt` chunks. | Magic: `89 50 4E 47 0D 0A 1A 0A` | Spec JSON: https://fileformat.co/api/v1/ext/png.json | Minimal Fixture (83 B, sha256:275f56b84188ce56...): https://fileformat.co/fixtures/v1/png/minimal-2x2-rgba.png
- [.usdz (USDZ)](https://fileformat.co/ext/usdz): Zero-compression, 64-byte mmap-aligned PKZIP package created by Pixar & Apple for AR QuickLook and VisionOS spatial scenes, beginning with a root `.usdc` (`PXR-USDC`) or `.usda` stage. | Magic: `50 4B 03 04` | Spec JSON: https://fileformat.co/api/v1/ext/usdz.json | Minimal Fixture (206 B, sha256:84ed4c279742f4c9...): https://fileformat.co/fixtures/v1/usdz/minimal-64b-aligned.usdz
- [.webp (WebP)](https://fileformat.co/ext/webp): RIFF-based modern raster image container supporting lossy (`VP8 `), lossless (`VP8L`), and extended alpha/animated/EXIF (`VP8X`) bitstreams. | Magic: `52 49 46 46 ?? ?? ?? ?? 57 45 42 50` | Spec JSON: https://fileformat.co/api/v1/ext/webp.json | Minimal Fixture (58 B, sha256:297ca327517d1621...): https://fileformat.co/fixtures/v1/webp/minimal-vp8x.webp

### DevTools, Captures & Bundles
- [.aab (Android AAB)](https://fileformat.co/ext/aab): Google Play publishing format packaging `BundleConfig.pb`, Protocol-Buffer encoded `base/manifest/AndroidManifest.xml`, and DEX bytecode inside a modular ZIP container. | Magic: `50 4B 03 04` | Spec JSON: https://fileformat.co/api/v1/ext/aab.json | Minimal Fixture (450 B, sha256:2564e236b066e818...): https://fileformat.co/fixtures/v1/aab/minimal-base.aab
- [.har (HAR)](https://fileformat.co/ext/har): UTF-8 JSON network session capture exported by Chrome/Firefox/Playwright containing full HTTP headers, cookies, request payloads, and timing waterfalls. | Magic: `structural` | Spec JSON: https://fileformat.co/api/v1/ext/har.json | Minimal Fixture (1642 B, sha256:be133a81b716a13a...): https://fileformat.co/fixtures/v1/har/minimal.har
- [.ipa (iOS IPA)](https://fileformat.co/ext/ipa): ZIP-compressed iOS/iPadOS application distribution archive structured around `Payload/<AppName>.app/Info.plist` (`bplist00`) and Mach-O ARM64 binaries. | Magic: `50 4B 03 04` | Spec JSON: https://fileformat.co/api/v1/ext/ipa.json | Minimal Fixture (694 B, sha256:85778c5134a0d4c6...): https://fileformat.co/fixtures/v1/ipa/minimal-app.ipa
- [.pcap (PCAP)](https://fileformat.co/ext/pcap): Classic 24-byte header packet capture format produced by `tcpdump` and Wireshark, encoding host endianness and timestamp precision directly in its 4-byte magic number. | Magic: `D4 C3 B2 A1` | Spec JSON: https://fileformat.co/api/v1/ext/pcap.json | Minimal Fixture (82 B, sha256:8e6bf3f1ae8790e7...): https://fileformat.co/fixtures/v1/pcap/minimal-ethernet.pcap
- [.pdf (PDF)](https://fileformat.co/ext/pdf): Object-graph document format parsed bottom-up from its trailing `startxref` and `%%EOF` markers—primary battleground for AI RAG table extraction and active `/JavaScript` upload sanitization. | Magic: `25 50 44 46 2D` | Spec JSON: https://fileformat.co/api/v1/ext/pdf.json | Minimal Fixture (1376 B, sha256:3d510c3e24ce7b48...): https://fileformat.co/fixtures/v1/pdf/minimal-1page.pdf
- [.wasm (WASM)](https://fileformat.co/ext/wasm): Portable, sandboxed stack-machine binary bytecode format (`\0asm` v1) structured as a sequence of LEB128-framed Type, Import, Function, Memory, Export, and Code sections. | Magic: `00 61 73 6D 01 00 00 00` | Spec JSON: https://fileformat.co/api/v1/ext/wasm.json | Minimal Fixture (41 B, sha256:f61fd62f57c41269...): https://fileformat.co/fixtures/v1/wasm/minimal-add.wasm
- [.xlsx (XLSX)](https://fileformat.co/ext/xlsx): ZIP-packaged XML spreadsheet container (`[Content_Types].xml` + `xl/workbook.xml`). Frequently misdiagnosed when backend export endpoints fail and save HTML/JSON error pages as `.xlsx`. | Magic: `50 4B 03 04` | Spec JSON: https://fileformat.co/api/v1/ext/xlsx.json | Minimal Fixture (4843 B, sha256:872a54723dd14451...): https://fileformat.co/fixtures/v1/xlsx/minimal-sheet.xlsx