What you can connect¶
Today¶
- A folder on your computer —
limma sync ./folder --url https://api.limma.aisends every file Limma reads, and each later run sends only what changed. This covers Dropbox, OneDrive and Google Drive too: sync the folder their desktop app keeps on your disk. (A folder) - S3, Azure Blob and Google Cloud Storage buckets —
limma sync s3://bucket/prefix(oraz://…,gs://…), read with your own machine's cloud credentials; Limma stores none. (A bucket) - Google Drive and SharePoint (OneDrive for Business included) —
limma sync gdrive://<folder-id>orlimma sync sharepoint://<tenant>.sharepoint.com/sites/<site>/<library>, read-only; Google Docs, Sheets and Slides arrive as Word, Excel and PowerPoint. (Google Drive, SharePoint) - Files from an agent — your agent asks
limma_ingestfor a one-time upload link and runs the command it is given. (Files from an agent) - Exports of records — a CSV, TSV, JSON or JSON Lines, or Excel file from a CRM, an ERP or a directory: each row becomes a record your agent can search and count. (CSV and Excel exports)
pip install limma-ai installs the limma command. It runs on your machine and reads
your files with your credentials: nothing is set up on Limma's side.
File formats¶
| kind | file types |
|---|---|
.pdf — scanned pages are read by OCR, in English |
|
| Word | .docx, .doc |
| Excel | .xlsx, .xlsm, .xls — save an old .xlsb as .xlsx first |
| PowerPoint | .pptx — save an old .ppt as .pptx first |
| OpenDocument (LibreOffice) | .odt, .ods, .odp |
| CSV and TSV (each also read as rows: one relation a table) | .csv, .tsv, .tab |
| JSON | .json |
| Text and Markdown | .txt, .text, .log, .md, .markdown |
| HTML | .html, .htm |
.eml, .msg (Outlook) — attachments are read too |
|
| RTF | .rtf |
| XML and XBRL | .xml, .xbrl |
| Zip archives | .zip — every file in it is read by its own type |
| Images | .png, .jpg, .jpeg, .tif, .tiff, .bmp, .gif, .webp — read by OCR, in English |
A file of any other type is skipped, and the sync says which and why.
Not there yet¶
None of these has a connector yet. Each has a way in today:
| source | the way in today |
|---|---|
| Notion | export the workspace as Markdown or HTML, then sync the folder |
| Slack | export the workspace's messages (JSON files), then sync the folder |
| Confluence | export the space as HTML or PDF, then sync the folder |
| GitHub | clone the repository and sync it: its Markdown and text files are read |
| Gmail and Outlook | save the messages you need as .eml or .msg files into a synced folder |
| Jira | export the issues as CSV and load them as records |
| CRMs (Salesforce, HubSpot and others) | export the objects as CSV and load them as records |
| A database, on the hosted service | export the tables as CSV or JSON Lines and load them as records — inside your own network Limma reads PostgreSQL, MySQL and SQLite directly |
Need one of these sooner? Write to contact@limma.ai.
Legacy Excel (.xls) files are read with xlrd. This product includes software developed by David Giffin david@giffin.org.