Skip to content

Ingest files

Read this if you want OMem to ingest documents from a folder on your Mac — your synced OneDrive, an iCloud Drive folder, a project directory, anything. The file kind is the universal one: point it at a directory and it ingests every supported format inside.

The fastest path is the setup wizard, which auto-detects OneDrive / iCloud / Dropbox / Documents and lets you pick. To configure it by hand:

  1. Set the folders to ingest (the “roots”):

    Terminal window
    omem config set kinds.file.source_config.roots '["~/Library/CloudStorage/OneDrive-Personal", "~/Documents/projects"]'
  2. Enable the kind:

    Terminal window
    omem config set kinds.file.enabled true
  3. Ingest:

    Terminal window
    omem ingest --kind file

Every supported format inside your roots — PDF, Word, PowerPoint, Excel, Markdown, text, HTML, images. Two scope filters keep it sane (both adjustable):

SettingDefaultWhat it does
kinds.file.scope.max_file_size_mb50Skip files larger than this.
kinds.file.scope.exclude_patterns["~$*", ".DS_Store", "node_modules/**"]Glob patterns to skip.

For example, to also skip a tmp folder:

Terminal window
omem config set kinds.file.scope.exclude_patterns '["~$*", ".DS_Store", "node_modules/**", "tmp/**"]'
Terminal window
omem ingest status # what the last run did
omem wiki ls --kind file # the pages it produced

If something didn’t ingest, omem ingest errors shows what failed and why. See observing ingest.

When OMem walks a synced OneDrive folder and finds a .loop file, it doesn’t try to parse that binary — it hands it to the loop kind to resolve from SharePoint. See ingest Loop notes.