🤖 feat: support mux:// deep links for new chat drafts #2069
Conversation
Add a shared MuxDeepLinkPayload type and parseMuxDeepLink() helper for parsing mux://chat/new deep links (projectPath/projectId/prompt/sectionId), with unit tests.
In packaged Electron, Windows/Linux deep links are passed in argv. Treating `mux://...` as an Electron launch arg ensures the desktop app is opened instead of routing to CLI help.
Register mux:// protocol and forward deep links from main to renderer.
Bridge mux:deep-link IPC events through preload by buffering payloads until the renderer subscribes, and expose them on window.api. Also extend WindowApi typings with consumePendingDeepLinks() and onDeepLink().
Handle mux:// deep links by opening a fresh workspace draft for the resolved project and prefilling the creation input text (without auto-sending). Startup links that arrive before projects load are buffered in-memory and retried.
--- _Generated with `mux` • Model: `openai:gpt-5.2` • Thinking: `xhigh` • Cost: `.91`_ <!-- mux-attribution: model=openai:gpt-5.2 thinking=xhigh costs=1.91 -->
--- _Generated with `mux` • Model: `openai:gpt-5.2` • Thinking: `xhigh` • Cost: `$7.99`_ <!-- mux-attribution: model=openai:gpt-5.2 thinking=xhigh costs=7.99 -->
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: a957bf3272
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
|
@codex review |
Subscribe before draining deep-link buffers and only buffer when there are no subscribers. This avoids a startup race where mux:// deep links can be lost between consume + subscription. --- _Generated with `mux` • Model: `openai:gpt-5.2` • Thinking: `xhigh` • Cost: `$8.49`_ <!-- mux-attribution: model=openai:gpt-5.2 thinking=xhigh costs=8.49 -->
ibetitsmike
force-pushed
the
mike/deep-link-yc3k
branch
from
2efe775 to
42a73fc
Compare
|
@codex review Fixed the deep-link startup race by subscribing before draining the preload buffer, and updated preload to only buffer when there are no subscribers. |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 42a73fc7bb
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
Queue unresolved mux:// deep links when the API/projects list is not ready yet, and retry once projects load. This avoids dropping links during startup in the window where ProjectContext can report loading=false before the API connection is established. --- _Generated with `mux` • Model: `openai:gpt-5.2` • Thinking: `xhigh` • Cost: `$8.49`_ <!-- mux-attribution: model=openai:gpt-5.2 thinking=xhigh costs=8.49 -->
|
@codex review Handled startup project-loading race: unresolved mux:// deep links are now buffered when the API isn't connected yet / projects list isn't ready, and retried once projects load. |
|
Codex Review: Didn't find any major issues. You're on a roll. ℹ️ About Codex in GitHubCodex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback". |
1 participant
Summary
Adds desktop deep links via the custom
mux://protocol so a URL can launch/focus Mux, select a project, and open a new chat draft with the prompt prefilled (prefill-only; does not auto-send).Background
External tools (scripts, issue trackers, notifications) can now jump directly into Mux with context, without requiring users to manually find the project and paste an initial prompt.
Implementation
mux://chat/new?...) shared by main/preload/renderer.mux://...arg.electron-builderconfig and added main-process capture for:open-urlsecond-instanceargvwindow.api.consumePendingDeepLinks()/window.api.onDeepLink().projectPath/projectIdValidation
make static-checkRisks
open-urland argv/second-instancepaths and buffering until renderer is ready.Generated with
mux• Model:openai:gpt-5.2• Thinking:xhigh• Cost:$7.99