You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Split out of #101, which now covers the VS Code extension only.
Why separate
The two distributions in #101 read as "the same thing twice", but they share no code and almost no toolchain:
VS Code
JetBrains
Language
TypeScript
Kotlin / Java
Build
npm + vsce
Gradle + IntelliJ Platform Plugin SDK
Embedding
WebView
JCEF
Distribution
VS Code Marketplace (Azure DevOps publisher)
JetBrains Marketplace (separate account + review)
Target
one IDE
the whole IntelliJ family (IDEA, Rider, GoLand, PyCharm, …)
Bundling them meant one ticket could not finish until both ecosystems were stood up. Splitting lets the VS Code extension ship while this one is scoped properly.
Scope
Same shape as the VS Code extension once that lands: the plugin starts a local bowire process with the project root as its working directory and hosts the workbench in a JCEF panel. Collections, environments and recordings therefore live in the project's .bowire/ folder and travel with the repo — no IDE-proprietary storage.
Open questions
JCEF availability across the IntelliJ family and minimum platform version
The scope here is sound and the split from #101 was right — the two IDE distributions genuinely share no code. But the table in the issue is also the argument against starting: Kotlin/Java, Gradle and the IntelliJ Platform Plugin SDK, JCEF instead of a WebView, and a separate marketplace account with its own review process. That is a second toolchain to stand up, learn and keep current, for a plugin whose job is to host the same workbench the VS Code extension already hosts.
Bowire is better served right now by depth on what exists than by a second front. The workbench, the CLI and MCP are three surfaces already held to parity; a fourth distribution channel adds maintenance without adding capability.
Nothing here is wasted if it comes back: the shape is settled — start a local bowire with the project root as its working directory, host it in a JCEF panel, and let collections and environments live in the project's .bowire/ folder so they travel with the repo. That is the same contract the VS Code extension is built on, so whoever picks this up inherits a working design rather than a blank page.
Removed from the v2.6 milestone so the version's history reflects what shipped. Reopen when there is a reason to carry a second toolchain — a user asking for it by name would be a good one.
Activity
feat(vscode): host the workbench in VS Code (#101)
feat(storage): Collections koennen im Repo liegen statt im Benutzerpr…
thomas-stegemann commentedon Sep 2, 2026
Not planned for now.
The scope here is sound and the split from #101 was right — the two IDE distributions genuinely share no code. But the table in the issue is also the argument against starting: Kotlin/Java, Gradle and the IntelliJ Platform Plugin SDK, JCEF instead of a WebView, and a separate marketplace account with its own review process. That is a second toolchain to stand up, learn and keep current, for a plugin whose job is to host the same workbench the VS Code extension already hosts.
Bowire is better served right now by depth on what exists than by a second front. The workbench, the CLI and MCP are three surfaces already held to parity; a fourth distribution channel adds maintenance without adding capability.
Nothing here is wasted if it comes back: the shape is settled — start a local
bowirewith the project root as its working directory, host it in a JCEF panel, and let collections and environments live in the project's.bowire/folder so they travel with the repo. That is the same contract the VS Code extension is built on, so whoever picks this up inherits a working design rather than a blank page.Removed from the v2.6 milestone so the version's history reflects what shipped. Reopen when there is a reason to carry a second toolchain — a user asking for it by name would be a good one.