AI Diary commands
The diary namespace exposes daily entries only. It uses report_type=daily and has no weekly/monthly option, generation, retry, settings, or HTML output. All commands require ai_daily:read.
sharge diary list
sharge diary list <YYYY-MM>sharge diary list 2026-07 --jsondata is an array of entries with identifier (YYYYMMDD), nonnegative duration_seconds, extra.city, extra.keywords, extra.recording_count, title, nullable description, nullable cover thumbnail and large URLs, generated_at, and updated_at. Specify a real month; the CLI does not assume the current one. Repeating the read is safe.
sharge diary search
sharge diary search <keyword> [--limit <1..100>]limit defaults to 20; keywords must be 1–200 characters. Search results add matched_fields (title and/or body), matched_title, and matched_body_excerpt. data is an array. Repeating the search is safe.
sharge diary search "San Francisco" --limit 20 --jsonsharge diary get
sharge diary get <YYYYMMDD>sharge diary get 20260730 --jsonThe identifier must be a real date; 20260230 fails locally. Details include fixed report_type: "daily", identifier, status (waiting, queued, processing, success, failed, cancelled), timezone, period start/end, title, description, summary, duration, Markdown body, word count, and generation/update times. Details can exist before a body does, in which case markdown and word_count are null. Repeating the read is safe.
Unsupported commands and fields
There are no weekly/monthly get, generate, retry, or settings commands. Results do not expose HTML, OSS keys, templates, rendering details, internal metadata, or retry information.