Michael Frank Martin

It looks like unasync.py wasn't run also. The _sync version of the public_search/api.py still has the old urls. If you're running in a Jupyter Notebook, for example, you might try...

FWIW, executing the findOrCreate with a map rather than a loop seems to cause the same problem -- i.e., that duplicate rows are not being identified. ``` Promise.map(update, function(row){ var...

Very helpful to know. So if I execute synchronously, can I get it to work consistently even without a non-unique index? Regardless, is there a better approach to this problem...

Confirming that this issue and the reported workaround still exist as of today with Claude Desktop 0.8.0 running on macOS Sequoia 15.3.1. Using the `claude_desktop_config.json` suggested in the walkthrough results...