Jmac
My feedback
4 results found
-
4 votes
An error occurred while saving the comment
Jmac
shared this idea
·
-
60 votes
An error occurred while saving the comment
Jmac
commented
+1, with a concrete daily pain point.
I develop in VS Code (plus OpenCode and other LLM tools), with everything versioned in Git. My current Lumo workflow is ****: every iteration means copying code back and forth between editor and Lumo, re-establishing context manually, and pushing each evolution separately. Over dozens of iterations, the copy-paste tax kills the momentum.
Two things would remove this friction:
1) An official VS Code extension — Lumo as a coding companion inside the editor: ask about the selection, refactor a function, explain an error, without leaving the IDE.
2) A "Lumo tuned" mode for code review — point it at a branch or a PR diff (Git-aware), get a structured review, iterate. Currently there's no privacy-first alternative to Copilot/Claude Code for this, and I deliberately don't send my code to providers that train on user data.
An API approach (as suggested in other comments) would also work if it enables community extensions — but official editor support would be the real game changer.
Jmac
supported this idea
·
-
55 votes
Jmac
supported this idea
·
-
1,099 votes
Jmac
supported this idea
·
An error occurred while saving the comment
Jmac
commented
Love that this is moving forward. One gap in the current Mail connector (Lumo Desktop): it's read/send only — no write actions. Please add the ability to move emails to folders/labels, mark as read, and create folders, so Lumo can actually triage an inbox ("file all Google Play receipts into my Games folder"). With user approval before each action, this stays privacy-first and turns Lumo into a real inbox assistant.
P.S. Happy to file this as a separate feature idea if the team prefers dedicated threads for specific scope requests — let me know what works best.
Adding concrete agent-style scenarios to illustrate the request, since this is how I'd use the connector daily:
1. DAILY BRIEFING: "Summarize what arrived yesterday" (works today), then "file the newsletters, mark the invoices as unread, draft replies" — impossible today, done by hand in a second window.
2. ROUTING: "Move all my purchase receipts from the last two years into their dedicated folder" — conversational triage that follows my intent of the moment, which filters can't anticipate.
3. CLEANUP CAMPAIGNS: "List newsletters I haven't opened in a year, propose deleting them" → I review → approved batch goes to Trash. Assistant proposes, I decide, it executes.
4. FOLLOW-UP AWARENESS: "Which threads am I still waiting on a reply to?"
5. PROJECT WORK: "Gather everything on this topic from last month, label it, move it to the archive folder, summarize the thread."
Key point: the combination of a capable email agent + rigorous per-action permission settings is something no US-based assistant offers, because their business model conflicts with it. Proton is uniquely positioned to build this — privacy-first by design, locally via Bridge, with the user in control.