mm peek¶

Surface locally-extracted file metadata — dimensions, EXIF, codec, duration, content hash — directly from the file, without any LLM calls or cache reads.

Use peek for "what is this file?". Use cat for "what does this file say?".

Synopsis¶

mm peek FILE [FILE ...] [OPTIONS]

Options¶

Flag	Short	Type	Description
`--full`		flag	Include all document metadata fields (author, title, subject, keywords, creator, producer, pages). Off by default.
`--format FORMAT`	`-f`	enum	Output format: `rich`, `json`, `pretty-json`, `tsv`, `csv`

Metadata fields¶

Fields populated depend on the file kind. All formats emit the same flat column set; fields not applicable to a given kind are null.

Universal fields¶

Field	Description
`name`	Filename
`kind`	File kind: `image`, `video`, `audio`, `document`, `code`, `text`, `data`, `config`, `other`
`size`	File size in bytes
`mime`	MIME type inferred from extension
`magic_mime`	MIME type detected by content inspection (magika). Shown only when it differs from `mime`.
`content_hash`	xxh3 content hash (hex)

Image fields¶

Field	Description
`dimensions`	`WxH` pixel dimensions
`phash`	Perceptual hash (16-digit hex) for similarity detection
`exif_camera`	Camera make and model
`exif_date`	Date/time original from EXIF
`exif_gps`	GPS coordinates (lat, lon)
`exif_orientation`	EXIF orientation tag

Video fields¶

Field	Description
`dimensions`	`WxH` pixel resolution
`duration_s`	Duration in seconds
`fps`	Frames per second
`video_codec`	Video codec (e.g. `h264`, `av1`)
`audio_codec`	Audio codec (e.g. `aac`, `opus`)
`has_audio`	Whether an audio track is present

Audio fields¶

Field	Description
`duration_s`	Duration in seconds
`audio_codec`	Audio codec

Document fields (always shown)¶

Field	Description
`mime`	MIME type
`content_hash`	Content hash

Document fields (`--full` only)¶

Field	Description
`pages`	Page count
`doc_author`	Document author
`doc_title`	Document title
`doc_subject`	Document subject
`doc_keywords`	Document keywords
`doc_creator`	Creating application
`doc_producer`	PDF producer

Examples¶

# basic image metadata
mm peek photo.png

# video metadata
mm peek clip.mp4

# audio metadata
mm peek recording.mp3

# PDF with full document metadata
mm peek paper.pdf --full

# multi-file inspection
mm peek *.png --format tsv

# multi-file JSON output
mm peek photo.png clip.mp4 paper.pdf --format json

# pipe-friendly TSV
mm peek photo.png --format tsv

Rich panel output¶

In TTY mode the default output is a Rich panel per file, showing only populated fields. None fields are hidden to keep output dense and scannable.

A typical image panel:

╭─── photo.jpg ───╮
│ kind       image │
│ size       2.4 MB│
│ mime   image/jpeg│
│ hash   a3f2...   │
│ dimensions 4032x3024 │
│ camera  iPhone 15 Pro │
│ date  2024-03-15 14:22 │
│ gps  37.7749° N, 122.4194° W │
╰──────────────────╯

Performance¶

peek is always a direct, fresh scan — no cache reads, no database. Every invocation re-reads the file.

Video metadata uses native Rust parsers (mp4parse for MP4/MOV, matroska for MKV/WebM) — no ffmpeg dependency for peek. Duration, codec, and resolution are extracted in under 100 ms.

Notes¶

peek accepts paths from stdin when piped (one path per line).
--full only adds document-specific fields; it has no effect on images, video, or audio.
magic_mime is displayed only when it differs from the extension-inferred mime — a discrepancy indicates a misnamed or modified file.
phash is available for images only. Displayed as a 16-digit hex string.
GPS coordinates are shown when present in EXIF; many photos from modern smartphones include them.