API reference
1 endpoint · Overview
Every word must match the name, a non-core sidecar (tags, AI prompts, transcripts, media facts) or a live comment; any part of a word matches, case-insensitively. Name matches rank first, then closer names, then newer files. When the server has an embeddings model (SIDESPACE_EMBED_URL, ADR 0018), files close in meaning to the query follow, with matched ["meaning"] and a score, so a plain description finds files that share none of its words. Hits in a transcript or a timed comment, and pieces of a transcript close in meaning, are listed as moments. Only spaces the caller is a member of are searched.
qstringrequiredspacestringtypestringvideoimageaudiofolderothermodestringautokeywordmeaninglimitinteger200itemsSearchHit[]requiredidstringrequiredspace_idstringrequiredparent_idstring | nullrequirednamestringrequiredfiledirhead_version_idstring | nullrequiredsizeintegerrequiredmimestring | nullrequiredmeta_revintegerrequireddeletedbooleanrequiredspace_namestringrequiredfolderstringrequiredmatchedstring[]requiredsnippetstringmomentsSearchMoment[]at_msinteger (int64)requireduntil_msinteger (int64)textstringrequiredsourcestringrequiredtranscriptcommentsmeaningscorenumbercurl "https://side.example.com/v1/search?q=harbor%20night" \
-H "Authorization: Bearer $SIDESPACE_TOKEN"{
"items": [
{
"node": {},
"space_name": "space_name",
"folder": "folder",
"matched": [],
"snippet": "snippet",
"moments": [],
"score": 0.5
}
]
}