Library and metadata tools
These tools browse and describe your library without a search query. They filter, look up, and summarize.
query_media
Section titled “query_media”Filter the library by the details your photos carry: date, place, camera, dimensions, file type, and more. Where search_photos ranks by what is in the picture, query_media filters by hard facts.
Inputs are a set of optional filters, combined together. They include date ranges, a location box, camera make and model, image dimensions, file type, capture kind, whether the description contains transcribed text, and sort order. At least one filter is required; a call with no filters is rejected.
Capture kind is how the file was made: photograph (a camera photo), screenshot, export (a resized or exported copy), or unknown. Use capture_kind: photograph to get real photos and leave screenshots out. This is about how the file was made, not what it shows, so a photograph of a receipt is still a photograph.
Returns the matching photos, each with its file name, date, camera details, dimensions, a short description, and how the file was made. On the local connection, results also include each photo’s location on your device.
get_photo_details
Section titled “get_photo_details”Everything Avlok knows about a single photo.
Inputs
| Field | Type | Required |
|---|---|---|
media_id |
integer | yes |
Returns the full record: file name (and, on the local connection, its location on your device), date taken and other dates, location coordinates and altitude where available, full camera details (make, model, lens, focal length, aperture, ISO, shutter speed), image dimensions and file size, a short description, and, on the local connection, the spoken words for a transcribed video.
Fields that do not apply to a given photo come back empty rather than missing, so the shape is always the same. If the photo does not exist, the tool returns a clear “not found” message.
get_photo_file
Section titled “get_photo_file”Returns a thumbnail of the photo so the assistant can actually see it, not just read about it.
Inputs
| Field | Type | Required | Notes |
|---|---|---|---|
media_id |
integer | yes | |
include_path |
boolean | no | Also return the file’s location on your device (local connection only) |
Returns the thumbnail. On the local connection, the image itself is returned. On the cloud connection, the tool returns a signed link that expires within a day, which the result card uses to display the photo. Your full-resolution photo is never sent; the assistant sees the same small thumbnail Avlok uses everywhere.
list_recent_photos
Section titled “list_recent_photos”The most recent photos, newest first.
Inputs
| Field | Type | Required | Notes |
|---|---|---|---|
limit |
integer | no | How many to return |
Returns recent photos in the standard shape.
get_library_stats
Section titled “get_library_stats”A summary of what is in your library: how many photos, how many videos, how many have been described, how many videos have had their spoken words processed, date ranges, and similar high-level counts.
Inputs: none.
Returns the summary. Useful for an assistant to orient itself before it starts searching, or to answer questions like “how many photos do I have from last year.”