feat(desktop): provider-level toggle in model visibility dialog - #43492
feat(desktop): provider-level toggle in model visibility dialog#43492michalmielczynski wants to merge 3 commits into
Conversation
Add enableProvider/disableProvider/toggleProviderVisibility functions with tombstone support so disabled providers stay off across reloads. Toggle switch per provider header lets user disable entire cloud provider groups (Anthropic, Google, OpenAI) without auto-revert.
|
Competing fix for #43485 alongside #43496. #43496 is the focused fix (a provider:: sentinel to stop the hide-all snap-back); this PR bundles a provider-level toggle plus unrelated electron main/preload, titlebar, and thread-virtualizer changes. Marking as duplicate of the cleaner #43496 for the same root cause — maintainer can salvage the provider-toggle UX from here. |
|
Closing as duplicate per maintainer label. Provider-level visibility toggle is already being tracked in another PR in the queue. |
tonydwb
left a comment
There was a problem hiding this comment.
Code Review Summary
Verdict: Approved
Overview
Desktop feature PR adding provider-level toggle in the model visibility dialog. Allows users to hide/show all models from a given provider at the provider level rather than per-model.
Looks Good
- Clean UI feature with proper sentinel key handling
- Correctly inserts/removes emptyProviderSentinelKey when all models are hidden/restored
- Well-scoped changes (8 files)
- No security concerns or hardcoded secrets
Reviewed by Hermes Agent
Problem
When all models of a provider were toggled off, closing and reopening the
dialog would auto-re-enable defaults. Empty storage was treated as
"never customized", falling back to curated defaults.
Solution
slug::) — disabling all models stores a tombstone so`effectiveVisibleKeys` knows this was intentional.
Files changed
6/6 tests pass. Build succeeds.