You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This commit was created on GitHub.com and signed with GitHub’s verified signature.
Changed — may break callers
memo sync --ignore leaves a file or folder alone for that run. Notes already imported from it stay in Memo;
before, the run removed them and the next run imported them again. notes.ignore in the config still removes them.
Fixed
Project context, key-document commands and search key-document ordering accept an exact project name or alias,
as meeting searches do. Ambiguous names list candidates; numeric IDs, project:<id> and keys still work.
memo sync reports the files it could not read on every run, also in the text report. Each folder line counts
them (1 not read) and a line names each file and why, such as a file over 50 MiB or an .xls. Before, only the
first run's --json listed them, and the count of notes included them.
A notes folder that is gone is a normal error, not_found with the way to fix it, not "this is a bug (ENOENT)". memo doctor names the same fix in place of a bare memo folders.
memo notes show names a link that leads nowhere by its text, in place of an empty entry after a comma.
The hint for search by meaning gives a command that works without installing tg: npx -y @wirecat/tg-cli local models text download e5-small.
The docs: .doc files are read and need nothing extra (PDF and Word readers come with Memo); install leads with memo skill install --for all and explains npm 12's "install scripts blocked"; Himalaya 2.x and its config file;
the cache folder and MEMO_AUTO; plain memo tasks list leaves out project tasks; tg and max commands show a
profile; the README example is real output.