{
  "schema_version": "mentaview.public-ingestion-format-catalog/v1",
  "updated": "2026-09-03",
  "generated_from": "crates/multimodal-ingestion/src/universal.rs",
  "contract_version": "mentaview.universal-ingestion/v2",
  "registry_fingerprint_sha256": "04905b2a20448bb13cbc900dc59b46a57fcac1d436c7cc4d44806e1ec1bf14cb",
  "unique_extension_identifiers": 735,
  "registered_family_count": 23,
  "family_count_with_opaque_fallback": 24,
  "resolution_policy": "Suffix lists are resolved in registry order, then reconciled with declared media type, magic bytes and container markers. A suffix is never trusted by itself.",
  "truth_states": [
    {
      "id": "recognized",
      "meaning": "Signature, container, declaration and suffix were reconciled."
    },
    {
      "id": "planned",
      "meaning": "A bounded preservation DAG exists; no extraction is implied."
    },
    {
      "id": "configured",
      "meaning": "An exact operator-approved adapter, binary or model is available."
    },
    {
      "id": "extracted",
      "meaning": "Canonical evidence and stage receipts exist for this source checksum."
    },
    {
      "id": "qualified",
      "meaning": "That named format/tool path passed a pinned corpus and promotion gate."
    }
  ],
  "execution_modes": [
    {
      "id": "immediate",
      "label": "Immediate",
      "trigger": "Healthy first-pass evidence",
      "behavior": "Publish the measured result without creating background recovery work."
    },
    {
      "id": "bounded-sync",
      "label": "Bounded synchronous",
      "trigger": "Moderate, repairable degradation",
      "behavior": "Apply the configured escalation inside the request budget and retain partial evidence."
    },
    {
      "id": "durable-background",
      "label": "Durable background",
      "trigger": "Critical missing, multiply critical or structurally sparse evidence",
      "behavior": "Journal a checksum-bound job, return HTTP 202 with a scoped receipt, retry within budget and re-measure before promotion."
    },
    {
      "id": "lazy",
      "label": "Lazy / on demand",
      "trigger": "Query-specific depth is not yet justified",
      "behavior": "Keep archive members, dense frames, rows or objects unmaterialized until an authorized request activates them."
    },
    {
      "id": "parallel",
      "label": "Parallel branches",
      "trigger": "Independent meaning-bearing channels",
      "behavior": "Run within an explicit concurrency bound; one optional failure does not erase successful sibling evidence."
    }
  ],
  "qualified_paths": [
    {
      "format": "PDF",
      "status": "Qualified canary path",
      "evidence": "Poppler layout text and page rendering, with bounded Tesseract OCR geometry when configured."
    },
    {
      "format": "DOCX",
      "status": "Qualified canary path",
      "evidence": "OOXML recognition, isolated LibreOffice PDF derivation, then the qualified PDF evidence path."
    },
    {
      "format": "TIFF",
      "status": "Qualified canary path",
      "evidence": "Authenticated raster normalization, OCR boxes/confidence and privacy-filtered metadata."
    },
    {
      "format": "MOV",
      "status": "Qualified metadata path",
      "evidence": "QuickTime identity, video media type and duration metadata; general frame understanding remains adapter-gated."
    },
    {
      "format": "WAV / PCM16",
      "status": "Qualified probe path",
      "evidence": "WAV identity, duration metadata and model-free RMS, peak, silence, clipping and DC-offset measurements."
    },
    {
      "format": "PCM16 meeting recovery",
      "status": "Qualified narrow path",
      "evidence": "Three checksum-bound overlapping windows, local ASR and deterministic time reconciliation passed on the named 90-second AMI close-talk and far-field conditions; this is not general meeting accuracy."
    },
    {
      "format": "PCM16 VAD + pre-roll",
      "status": "Qualified narrow path",
      "evidence": "First-party adaptive energy gating skipped known silence and preserved absolute source time on the named Vosk short recording and AMI mixed-headset segment; this is not frame-labelled VAD accuracy or noisy voice-note qualification."
    },
    {
      "format": "FLAC / M4A-AAC / Ogg-Opus deferred recovery",
      "status": "Qualified narrow path",
      "evidence": "The exact signed-source FFmpeg 9.0.1 LGPL-only recipe decoded one real source to mono 16 kHz PCM16; healthy signals stayed byte-identical and one controlled quiet M4A received bounded x8 level repair before VAD."
    },
    {
      "format": "Real-room noisy speech lazy recovery",
      "status": "Qualified narrow path",
      "evidence": "Eight post-screening VOiCES room/babble/music/television recordings exercised sparse non-empty transcript detection and VAD-windowed retry at 12.93% micro WER for the exact profile; this is not representative field voice-note qualification."
    },
    {
      "format": "Prospective conversation quality fallback",
      "status": "Qualified narrow path",
      "evidence": "One frozen five-minute CHiME-6 four-speaker kitchen conversation with 34.00% overlap triggered continuous background retry; WER improved from 67.62% to 55.28% and coverage from 75.28% to 93.12%. Diarization did not run."
    },
    {
      "format": "Speaker timeline and diarization adapter",
      "status": "Contract and integration passed; model rejected",
      "evidence": "The provider-neutral audio v2 contract and real S21 loopback preserved 79 bounded spans, four anonymous clusters and overlap. Missed transcript spans remain unattributed without discarding a valid timeline or inventing a label. Frozen S01 and S21 model-quality gates failed, so the exact external profile remains experimental and unqualified."
    },
    {
      "format": "Real lazy speaker-timeline lifecycle",
      "status": "Qualified narrow lifecycle; model rejected",
      "evidence": "One frozen five-minute CHiME-6 S01 source survived an injected initial 503 with ASR preserved, completed one durable worker attempt and one successful complete-source diarization request, validated 85 records, persisted/read 76 spans and deleted the raw recovery source. This is single-process lifecycle evidence, not model-quality or production evidence."
    },
    {
      "format": "Weak-binarization TIFF recovery",
      "status": "Qualified narrow path",
      "evidence": "One checksum-pinned NOD historical page forced the durable quality-recovery v5 raster branch: 4/4 overlapping tiles completed with no failure, the raw recovery copy was deleted and reference CER fell from 16.12% to 3.23% (79.93% relative). This is not general OCR qualification."
    },
    {
      "format": "ZIP",
      "status": "Qualified bounded path",
      "evidence": "Safe hierarchy inventory, exact lazy member selection and authenticated two-level nested traversal under one aggregate budget."
    }
  ],
  "duplicate_declarations": [
    {
      "extension": "ts",
      "declared_by": [
        "source-code",
        "video"
      ],
      "resolved_to": "source-code"
    },
    {
      "extension": "mts",
      "declared_by": [
        "source-code",
        "video"
      ],
      "resolved_to": "source-code"
    },
    {
      "extension": "asm",
      "declared_by": [
        "source-code",
        "cad"
      ],
      "resolved_to": "source-code"
    },
    {
      "extension": "vhd",
      "declared_by": [
        "source-code",
        "disk-image"
      ],
      "resolved_to": "source-code"
    },
    {
      "extension": "ics",
      "declared_by": [
        "structured-data",
        "email"
      ],
      "resolved_to": "structured-data"
    },
    {
      "extension": "vcf",
      "declared_by": [
        "structured-data",
        "email",
        "scientific"
      ],
      "resolved_to": "structured-data"
    },
    {
      "extension": "pdb",
      "declared_by": [
        "ebook",
        "scientific"
      ],
      "resolved_to": "ebook"
    },
    {
      "extension": "hdr",
      "declared_by": [
        "raster-image",
        "medical"
      ],
      "resolved_to": "raster-image"
    },
    {
      "extension": "raw",
      "declared_by": [
        "raster-image",
        "disk-image"
      ],
      "resolved_to": "raster-image"
    },
    {
      "extension": "dxf",
      "declared_by": [
        "vector-image",
        "cad"
      ],
      "resolved_to": "vector-image"
    },
    {
      "extension": "ogg",
      "declared_by": [
        "audio",
        "video"
      ],
      "resolved_to": "audio"
    },
    {
      "extension": "rm",
      "declared_by": [
        "audio",
        "video"
      ],
      "resolved_to": "audio"
    },
    {
      "extension": "deb",
      "declared_by": [
        "archive",
        "executable"
      ],
      "resolved_to": "archive"
    },
    {
      "extension": "rpm",
      "declared_by": [
        "archive",
        "executable"
      ],
      "resolved_to": "archive"
    },
    {
      "extension": "apk",
      "declared_by": [
        "archive",
        "executable"
      ],
      "resolved_to": "archive"
    },
    {
      "extension": "ipa",
      "declared_by": [
        "archive",
        "executable"
      ],
      "resolved_to": "archive"
    },
    {
      "extension": "jar",
      "declared_by": [
        "archive",
        "executable"
      ],
      "resolved_to": "archive"
    },
    {
      "extension": "bdf",
      "declared_by": [
        "font",
        "medical"
      ],
      "resolved_to": "font"
    },
    {
      "extension": "obj",
      "declared_by": [
        "model-3d",
        "executable"
      ],
      "resolved_to": "model-3d"
    },
    {
      "extension": "ply",
      "declared_by": [
        "model-3d",
        "geospatial"
      ],
      "resolved_to": "model-3d"
    },
    {
      "extension": "fit",
      "declared_by": [
        "geospatial",
        "scientific"
      ],
      "resolved_to": "geospatial"
    },
    {
      "extension": "img",
      "declared_by": [
        "medical",
        "disk-image"
      ],
      "resolved_to": "medical"
    },
    {
      "extension": "mdf",
      "declared_by": [
        "database",
        "disk-image"
      ],
      "resolved_to": "database"
    },
    {
      "extension": "bin",
      "declared_by": [
        "disk-image",
        "executable"
      ],
      "resolved_to": "disk-image"
    }
  ],
  "families": [
    {
      "id": "text",
      "label": "Text and markup",
      "kind": "registered",
      "identifier_count": 19,
      "extensions": [
        "txt",
        "text",
        "md",
        "markdown",
        "mdown",
        "mkd",
        "rst",
        "adoc",
        "asciidoc",
        "org",
        "tex",
        "latex",
        "bib",
        "log",
        "nfo",
        "man",
        "me",
        "rtx",
        "readme"
      ],
      "icon": "/vendor/file-icons/Default.svg",
      "availability": "Native-first",
      "preserves": "Text, encoding, headings, paragraph order and source metadata."
    },
    {
      "id": "source-code",
      "label": "Source code",
      "kind": "registered",
      "identifier_count": 83,
      "extensions": [
        "c",
        "h",
        "cc",
        "cpp",
        "cxx",
        "hpp",
        "hxx",
        "m",
        "mm",
        "rs",
        "go",
        "py",
        "pyi",
        "pyx",
        "rb",
        "php",
        "java",
        "kt",
        "kts",
        "scala",
        "sc",
        "swift",
        "dart",
        "js",
        "mjs",
        "cjs",
        "jsx",
        "ts",
        "mts",
        "cts",
        "tsx",
        "vue",
        "svelte",
        "astro",
        "cs",
        "fs",
        "fsx",
        "vb",
        "vbs",
        "lua",
        "r",
        "jl",
        "pl",
        "pm",
        "t",
        "sh",
        "bash",
        "zsh",
        "fish",
        "ps1",
        "bat",
        "cmd",
        "sql",
        "graphql",
        "gql",
        "proto",
        "sol",
        "vy",
        "asm",
        "s",
        "v",
        "vhd",
        "vhdl",
        "clj",
        "cljs",
        "cljc",
        "edn",
        "ex",
        "exs",
        "erl",
        "hrl",
        "hs",
        "lhs",
        "ml",
        "mli",
        "nim",
        "zig",
        "groovy",
        "gradle",
        "cmake",
        "make",
        "mk",
        "dockerfile"
      ],
      "icon": "/vendor/file-icons/WebAssembly.svg",
      "availability": "Native-first",
      "preserves": "Text, language identity, line structure and code-semantics planning."
    },
    {
      "id": "structured-data",
      "label": "Structured data",
      "kind": "registered",
      "identifier_count": 41,
      "extensions": [
        "csv",
        "tsv",
        "psv",
        "ssv",
        "json",
        "jsonl",
        "ndjson",
        "json5",
        "jsonld",
        "geojson",
        "topojson",
        "xml",
        "xsd",
        "xsl",
        "xslt",
        "yaml",
        "yml",
        "toml",
        "ini",
        "cfg",
        "conf",
        "properties",
        "env",
        "plist",
        "opml",
        "rss",
        "atom",
        "ics",
        "vcf",
        "har",
        "sarif",
        "parquet",
        "avro",
        "orc",
        "arrow",
        "feather",
        "sav",
        "sas7bdat",
        "dta",
        "por",
        "dbf"
      ],
      "icon": "/vendor/file-icons/JSON-1.svg",
      "availability": "Native + adapter-gated",
      "preserves": "Records, fields, types, schemas, relationships and source metadata."
    },
    {
      "id": "web",
      "label": "Web and captured pages",
      "kind": "registered",
      "identifier_count": 12,
      "extensions": [
        "html",
        "htm",
        "xhtml",
        "mhtml",
        "mht",
        "css",
        "scss",
        "sass",
        "less",
        "webarchive",
        "warc",
        "warc.gz"
      ],
      "icon": "/vendor/file-icons/Default.svg",
      "availability": "Adapter-gated",
      "preserves": "DOM structure, reading order, styles, hyperlinks, embedded assets and capture metadata."
    },
    {
      "id": "document",
      "label": "Documents",
      "kind": "registered",
      "identifier_count": 32,
      "extensions": [
        "pdf",
        "pdfa",
        "xps",
        "oxps",
        "djvu",
        "djv",
        "rtf",
        "doc",
        "dot",
        "docx",
        "docm",
        "dotx",
        "dotm",
        "odt",
        "ott",
        "fodt",
        "wpd",
        "wps",
        "abw",
        "pages",
        "hwp",
        "hwpx",
        "sxw",
        "stw",
        "uot",
        "lwp",
        "cwk",
        "pub",
        "one",
        "onepkg",
        "chm",
        "hlp"
      ],
      "icon": "/vendor/file-icons/Default.svg",
      "availability": "Named paths qualified",
      "preserves": "Text, pages, reading order, geometry, tables, formulas, figures, comments and revisions."
    },
    {
      "id": "spreadsheet",
      "label": "Spreadsheets",
      "kind": "registered",
      "identifier_count": 27,
      "extensions": [
        "xls",
        "xlt",
        "xla",
        "xlsx",
        "xlsm",
        "xlsb",
        "xltx",
        "xltm",
        "xlam",
        "ods",
        "ots",
        "fods",
        "numbers",
        "sxc",
        "stc",
        "uos",
        "dif",
        "slk",
        "wk1",
        "wk2",
        "wk3",
        "wk4",
        "wks",
        "123",
        "gnumeric",
        "qpw",
        "et"
      ],
      "icon": "/vendor/file-icons/Default.svg",
      "availability": "Adapter-gated",
      "preserves": "Sheets, cells, formulas, tables, charts, types, comments and calculation relationships."
    },
    {
      "id": "presentation",
      "label": "Presentations",
      "kind": "registered",
      "identifier_count": 18,
      "extensions": [
        "ppt",
        "pps",
        "pot",
        "pptx",
        "pptm",
        "ppsx",
        "ppsm",
        "potx",
        "potm",
        "odp",
        "otp",
        "fodp",
        "key",
        "sxi",
        "sti",
        "uop",
        "show",
        "dps"
      ],
      "icon": "/vendor/file-icons/Default.svg",
      "availability": "Adapter-gated",
      "preserves": "Slides, spatial layout, notes, reading order, charts, media and transitions."
    },
    {
      "id": "ebook",
      "label": "Ebooks and comics",
      "kind": "registered",
      "identifier_count": 16,
      "extensions": [
        "epub",
        "mobi",
        "azw",
        "azw3",
        "kf8",
        "kfx",
        "fb2",
        "lit",
        "lrf",
        "pdb",
        "prc",
        "cbz",
        "cbr",
        "cb7",
        "cbt",
        "ibooks"
      ],
      "icon": "/vendor/file-icons/Default.svg",
      "availability": "Adapter-gated",
      "preserves": "Chapters, spine order, text, pagination, images, navigation and publication metadata."
    },
    {
      "id": "email",
      "label": "Email and calendars",
      "kind": "registered",
      "identifier_count": 8,
      "extensions": [
        "eml",
        "emlx",
        "msg",
        "mbox",
        "pst",
        "ost",
        "oft",
        "maildir"
      ],
      "icon": "/vendor/file-icons/Default.svg",
      "availability": "Adapter-gated",
      "preserves": "Headers, bodies, threads, attachments, recipients, dates and privacy-sensitive metadata."
    },
    {
      "id": "raster-image",
      "label": "Raster images",
      "kind": "registered",
      "identifier_count": 78,
      "extensions": [
        "jpg",
        "jpeg",
        "jpe",
        "jfif",
        "jif",
        "png",
        "apng",
        "gif",
        "bmp",
        "dib",
        "tif",
        "tiff",
        "webp",
        "avif",
        "heic",
        "heif",
        "jxl",
        "jp2",
        "j2k",
        "j2c",
        "jpf",
        "jpx",
        "jpm",
        "mj2",
        "ppm",
        "pgm",
        "pbm",
        "pnm",
        "pam",
        "pcx",
        "tga",
        "icb",
        "vda",
        "vst",
        "qoi",
        "ico",
        "cur",
        "ani",
        "icns",
        "dds",
        "ktx",
        "ktx2",
        "astc",
        "hdr",
        "rgbe",
        "exr",
        "psd",
        "psb",
        "xcf",
        "kra",
        "ora",
        "clip",
        "sai",
        "raw",
        "dng",
        "cr2",
        "cr3",
        "nef",
        "nrw",
        "arw",
        "srf",
        "sr2",
        "orf",
        "rw2",
        "raf",
        "pef",
        "x3f",
        "erf",
        "mrw",
        "3fr",
        "fff",
        "mos",
        "iiq",
        "rwl",
        "srw",
        "gpr",
        "bpg",
        "flif"
      ],
      "icon": "/vendor/file-icons/Image.svg",
      "availability": "Named paths qualified",
      "preserves": "Pixels, frames, orientation, OCR geometry, visual semantics, color and safe metadata."
    },
    {
      "id": "vector-image",
      "label": "Vector images and diagrams",
      "kind": "registered",
      "identifier_count": 23,
      "extensions": [
        "svg",
        "svgz",
        "eps",
        "epsf",
        "ai",
        "cdr",
        "cmx",
        "wmf",
        "emf",
        "sk",
        "sk1",
        "vsd",
        "vdx",
        "vsdx",
        "vssx",
        "vstx",
        "dia",
        "drawio",
        "fig",
        "odg",
        "fodg",
        "sxd",
        "dxf"
      ],
      "icon": "/vendor/file-icons/Image.svg",
      "availability": "Adapter-gated",
      "preserves": "Shapes, layers, text, coordinates, links, styling and rendered visual evidence."
    },
    {
      "id": "audio",
      "label": "Audio, voice and meetings",
      "kind": "registered",
      "identifier_count": 53,
      "extensions": [
        "wav",
        "wave",
        "bwf",
        "mp3",
        "mp2",
        "mp1",
        "aac",
        "m4a",
        "m4b",
        "m4p",
        "alac",
        "flac",
        "oga",
        "ogg",
        "opus",
        "vorbis",
        "wma",
        "aiff",
        "aif",
        "aifc",
        "au",
        "snd",
        "amr",
        "awb",
        "ac3",
        "eac3",
        "dts",
        "dtshd",
        "mka",
        "ape",
        "wv",
        "tta",
        "tak",
        "spx",
        "gsm",
        "caf",
        "ra",
        "rm",
        "mid",
        "midi",
        "kar",
        "rmi",
        "mod",
        "xm",
        "it",
        "s3m",
        "mptm",
        "669",
        "voc",
        "aa",
        "aax",
        "dsf",
        "dff"
      ],
      "icon": "/vendor/file-icons/WebVTT.svg",
      "availability": "Named ASR paths qualified; diarization contract integrated, model rejected",
      "preserves": "Tracks, signal measurements, speech, timestamps, speakers, audio events and provenance."
    },
    {
      "id": "video",
      "label": "Video",
      "kind": "registered",
      "identifier_count": 42,
      "extensions": [
        "mp4",
        "m4v",
        "mov",
        "qt",
        "webm",
        "mkv",
        "mk3d",
        "avi",
        "divx",
        "wmv",
        "asf",
        "flv",
        "f4v",
        "f4p",
        "f4a",
        "f4b",
        "mpeg",
        "mpg",
        "mpe",
        "m1v",
        "m2v",
        "mpv",
        "m2ts",
        "m2t",
        "vob",
        "ogv",
        "3gp",
        "3g2",
        "3gpp",
        "3gpp2",
        "rmvb",
        "dv",
        "dvr-ms",
        "wtv",
        "mxf",
        "roq",
        "nsv",
        "y4m",
        "h264",
        "h265",
        "hevc",
        "av1"
      ],
      "icon": "/vendor/file-icons/Video.svg",
      "availability": "Metadata path qualified",
      "preserves": "Tracks, time, speech, scenes, keyframes, frame OCR, visual semantics and metadata."
    },
    {
      "id": "archive",
      "label": "Archives and packages",
      "kind": "registered",
      "identifier_count": 56,
      "extensions": [
        "zip",
        "zipx",
        "tar",
        "tgz",
        "tar.gz",
        "tbz",
        "tbz2",
        "tar.bz2",
        "txz",
        "tar.xz",
        "tlz",
        "tar.lz",
        "tzst",
        "tar.zst",
        "gz",
        "gzip",
        "bz2",
        "bzip2",
        "xz",
        "lz",
        "lzma",
        "zst",
        "z",
        "7z",
        "rar",
        "cab",
        "arj",
        "lha",
        "lzh",
        "ace",
        "arc",
        "cpio",
        "shar",
        "sit",
        "sitx",
        "sea",
        "pak",
        "wim",
        "swm",
        "xar",
        "ar",
        "deb",
        "rpm",
        "apk",
        "ipa",
        "jar",
        "war",
        "ear",
        "whl",
        "crate",
        "gem",
        "nupkg",
        "vsix",
        "appx",
        "msix",
        "unitypackage"
      ],
      "icon": "/vendor/file-icons/Brotli.svg",
      "availability": "Named ZIP paths qualified",
      "preserves": "Hierarchy, member identity, checksums, compression risk and selected nested materialization."
    },
    {
      "id": "font",
      "label": "Fonts",
      "kind": "registered",
      "identifier_count": 16,
      "extensions": [
        "ttf",
        "otf",
        "ttc",
        "otc",
        "woff",
        "woff2",
        "eot",
        "pfb",
        "pfm",
        "afm",
        "bdf",
        "pcf",
        "fon",
        "fnt",
        "suit",
        "dfont"
      ],
      "icon": "/vendor/file-icons/Font.svg",
      "availability": "Safe-static only",
      "preserves": "Signature, tables, names, glyph and variation metadata; fonts are never executed as code."
    },
    {
      "id": "model-3d",
      "label": "3D models",
      "kind": "registered",
      "identifier_count": 34,
      "extensions": [
        "gltf",
        "glb",
        "obj",
        "fbx",
        "dae",
        "3ds",
        "blend",
        "ply",
        "stl",
        "3mf",
        "amf",
        "x3d",
        "x3db",
        "x3dv",
        "wrl",
        "vrml",
        "usd",
        "usda",
        "usdc",
        "usdz",
        "abc",
        "bvh",
        "lwo",
        "lws",
        "ma",
        "mb",
        "c4d",
        "max",
        "skp",
        "step",
        "stp",
        "iges",
        "igs",
        "brep"
      ],
      "icon": "/vendor/file-icons/3D-Model.svg",
      "availability": "Adapter-gated",
      "preserves": "Scene graph, geometry, materials, textures, cameras, animation and rendered views."
    },
    {
      "id": "cad",
      "label": "CAD and fabrication",
      "kind": "registered",
      "identifier_count": 24,
      "extensions": [
        "dwg",
        "dws",
        "dwt",
        "ifc",
        "ifczip",
        "rvt",
        "rfa",
        "sat",
        "sab",
        "prt",
        "catpart",
        "catproduct",
        "sldprt",
        "sldasm",
        "ipt",
        "iam",
        "fcstd",
        "sch",
        "brd",
        "kicad_sch",
        "kicad_pcb",
        "ger",
        "gbr",
        "excellon"
      ],
      "icon": "/vendor/file-icons/3D-Model.svg",
      "availability": "Adapter-gated",
      "preserves": "Assemblies, parts, layers, units, coordinates, dimensions and safe rendered views."
    },
    {
      "id": "geospatial",
      "label": "Geospatial",
      "kind": "registered",
      "identifier_count": 32,
      "extensions": [
        "shp",
        "shx",
        "prj",
        "cpg",
        "gpkg",
        "kml",
        "kmz",
        "gml",
        "gpx",
        "fit",
        "tcx",
        "osm",
        "pbf",
        "mbtiles",
        "pmtiles",
        "geotiff",
        "cog",
        "asc",
        "grd",
        "dem",
        "bil",
        "bip",
        "bsq",
        "las",
        "laz",
        "e57",
        "vrt",
        "sdat",
        "qgz",
        "qgs",
        "mxd",
        "lyr"
      ],
      "icon": "/vendor/file-icons/Image.svg",
      "availability": "Adapter-gated",
      "preserves": "Layers, coordinates, reference systems, features, attributes, rasters and provenance."
    },
    {
      "id": "medical",
      "label": "Medical and life-science",
      "kind": "registered",
      "identifier_count": 20,
      "extensions": [
        "dcm",
        "dicom",
        "nii",
        "nii.gz",
        "nrrd",
        "mhd",
        "mha",
        "mnc",
        "ima",
        "img",
        "edf",
        "fhir",
        "hl7",
        "cda",
        "svs",
        "ndpi",
        "mrxs",
        "scn",
        "vms",
        "vmu"
      ],
      "icon": "/vendor/file-icons/Default.svg",
      "availability": "Adapter-gated",
      "preserves": "Study/series hierarchy, frames, measurements, structured fields and sensitive metadata controls."
    },
    {
      "id": "scientific",
      "label": "Scientific and research",
      "kind": "registered",
      "identifier_count": 43,
      "extensions": [
        "h5",
        "hdf",
        "hdf5",
        "he5",
        "nc",
        "nc4",
        "cdf",
        "fits",
        "fts",
        "mat",
        "npz",
        "npy",
        "pkl",
        "pickle",
        "joblib",
        "rdata",
        "rds",
        "root",
        "vtk",
        "vti",
        "vtu",
        "mrc",
        "map",
        "dx",
        "cube",
        "cif",
        "mmcif",
        "mol",
        "mol2",
        "sdf",
        "xyz",
        "fasta",
        "fa",
        "fastq",
        "fq",
        "bam",
        "sam",
        "cram",
        "bcf",
        "bed",
        "gff",
        "gff3",
        "gtf"
      ],
      "icon": "/vendor/file-icons/JSON-1.svg",
      "availability": "Adapter-gated",
      "preserves": "Datasets, arrays, schemas, units, axes, provenance and domain-specific relationships."
    },
    {
      "id": "database",
      "label": "Databases",
      "kind": "registered",
      "identifier_count": 20,
      "extensions": [
        "db",
        "sqlite",
        "sqlite3",
        "db3",
        "mdb",
        "accdb",
        "fdb",
        "gdb",
        "nsf",
        "sqlitedb",
        "duckdb",
        "realm",
        "leveldb",
        "ldb",
        "sst",
        "rdb",
        "ndf",
        "mdf",
        "ldf",
        "bak"
      ],
      "icon": "/vendor/file-icons/SQLite.svg",
      "availability": "Adapter-gated",
      "preserves": "Schema, tables, indexes, relationships and explicitly selected bounded rows."
    },
    {
      "id": "disk-image",
      "label": "Disk and virtual-machine images",
      "kind": "registered",
      "identifier_count": 15,
      "extensions": [
        "iso",
        "dmg",
        "sparseimage",
        "sparsebundle",
        "vhdx",
        "vmdk",
        "vdi",
        "qcow",
        "qcow2",
        "bin",
        "cue",
        "nrg",
        "mds",
        "toast",
        "udf"
      ],
      "icon": "/vendor/file-icons/Default.svg",
      "availability": "Safe-static only",
      "preserves": "Container identity, partition/filesystem metadata and threat flags; never mounted automatically."
    },
    {
      "id": "executable",
      "label": "Executables and packages",
      "kind": "registered",
      "identifier_count": 23,
      "extensions": [
        "exe",
        "dll",
        "sys",
        "com",
        "msi",
        "msp",
        "scr",
        "cpl",
        "elf",
        "so",
        "dylib",
        "app",
        "appimage",
        "run",
        "wasm",
        "class",
        "dex",
        "o",
        "a",
        "lib",
        "ko",
        "efi",
        "rom"
      ],
      "icon": "/vendor/file-icons/WebAssembly.svg",
      "availability": "Safe-static only",
      "preserves": "Signature, platform, sections, imports, strings and package metadata; never executed."
    },
    {
      "id": "unknown-binary",
      "label": "Unknown binary",
      "kind": "opaque-fallback",
      "identifier_count": null,
      "extensions": [],
      "icon": "/vendor/file-icons/Default.svg",
      "availability": "Opaque fallback",
      "preserves": "Source checksum, size, declarations, threat flags and adapter-resolution plan; no semantic extraction claim."
    }
  ]
}