Hi, thanks for the report.
We recently found that the feature wasn’t working correctly - while the title of your old chat was inserted, the actual summary wasn’t, so the AI was guessing at the contents of your past chat when you referenced it.
While we may bring the feature back in a functional state in the future, we thought it best to simplify the context menu and remove features that were either not functional or not effective in serving their purpose.
As a workaround, using /summarize effectively summarizes your past chat, so you can continue working without significant context bloat. This happens automatically when the context approaches the model’s limit, but this command allows you to trigger it manually when it is best for your workflow.
Alternatively, if you lean more toward splitting your work into separate chats, you may want to create your own command that summarizes the thread into a few sentences, which you can copy and paste into a new chat for context.
You can learn more about that here: