Create (or overwrite) a Talk room; returns {url, room_name, conference_id, title, room}.

Schedule a calendar meeting; returns {url, event}.

Past meetings, newest first — deep paginated archive. Default row: conference_key, room_name, title, start, end, participants_count, participants (≤6; full list via get_meeting), has_chat, recording_ids. Times are ISO-8601 in the machine's own zone, to the second. Meetings still in progress may also appear (the request asks for them the way the web client does; not verified against a live meeting). Returns {meetings, count, has_more, next_offset} — the API gives no grand total, so has_more is the only end signal. A requested field the API did not send is simply missing from the row: that means "not in this payload", not "no data at all".

One meeting with its full participant list. Returns everything the payload has by default — detail is asked for by key, and for the detail: conference_key, room_name, title, start, end (ISO-8601, local zone, to the second), participants_count, participants and invited_participants (full; a guest without an account → {name, is_anonymous: true} — anonymousName is all the identification they have), has_chat, recording_ids, session_halls_artifacts. A field the API did not send is simply missing, which means "not in this payload", not "no data at all".

Recordings, newest first. Default row: recording_id, title, created_date (ISO-8601, local zone, to the second), created_by, duration, participants_count, participants (≤6; full list via get_recording), status, has_audio. There is no date window here — the recordings endpoint has no date filter, so narrow with query and limit instead. Returns {recordings, count, has_more, next_offset} — the API gives no grand total, so has_more is the only end signal. A requested field the API did not send is simply missing from the row: that means "not in this payload", not "no data at all".

Recording detail with its full participant list. Returns everything the payload has by default — detail is asked for by id, and for the detail. Surfaces conference_key (link to the meeting/chat; only on this endpoint), download_url (best-quality file for download_recording) and has_chat, alongside title, description, created_date (ISO-8601, local zone, to the second), created_by, duration, status, participants_count, participants, has_audio, comments_count, frame_size, preview_image, qualities. The last four are bulky — pass fields to leave them out. A field the API did not send is simply missing, i.e. "not in this payload".

Download a recording's video file to disk (best available quality). An existing file at target_path is overwritten.

Meeting chat across all channels, merged and time-ordered. The chat lives on the meeting, not the recording — pass conference_key, or a recording_id whose conference_key is resolved. With save_path, writes .json + .txt and returns a compact summary instead of the messages. Returns {conference_key, channels, count, truncated_channels, messages:[{id, channel, created, author, authorLogin, isAnonymous, text, files, reactions}]}; a non-empty truncated_channels means those channels hit limit and the chat may be incomplete.

Token status: authenticated/expired, expiry, the token-file path, and the README link. No parameters. To (re)authenticate, put a Session token in the token_file.