`omarchy plugin enable <id>` and `omarchy bar put <id>` both report success and place nothing when the widget's id is already listed in `config.plugins[]` but is not on the bar. The shell's own listing keeps reporting the plugin as disabled afterwards, so the two disagree and the user is told the opposite of what happened.

## Reproduce

Put a bar-widget's id in `plugins` in `~/.config/omarchy/shell.json` without placing it in `bar.layout`:

```json
"plugins": [{ "id": "agx.screen-time" }]
```

Then:

```
$ omarchy plugin enable agx.screen-time
Enabled agx.screen-time                     # exit 0

$ omarchy bar put agx.screen-time --section right
agx.screen-time is on the bar               # exit 0

$ omarchy-shell shell listPlugins | jq -r '.[]|select(.id=="agx.screen-time")|.enabled'
false
```

`shell.json` is byte-identical before and after both commands — neither wrote anything. The widget is not on the bar.

I hit this with three plugins in that state at once (`agx.screen-time`, `oma.nearby`, `omaconnect`, all `kinds: ["service", "bar-widget"]`). The only way out was editing `shell.json` by hand, because every command that is supposed to fix it claims it already did.

## Cause

`PluginRegistry.setEnabled` decides whether to place the widget with `findEntryLocation`:

https://github.com/omacom/omarchy/blob/quattro/shell/services/PluginRegistry.qml#L522

```js
var location = findEntryLocation(config, key)
...
if (!location.found && isBarWidget) {
  // ... place it in bar.layout
} else if (!location.found && !isFirstParty) {
  config.plugins.push(entry)
}
```

But `findEntryLocation` searches `bar.layout` **and** `config.plugins[]`:

https://github.com/omacom/omarchy/blob/quattro/shell/services/PluginRegistry.qml#L242-L246

So a bar-widget sitting in `plugins[]` reports `found: true`, both branches are skipped, and `setEnabled` still returns `true` — which the CLI prints as success.

The two questions have been conflated. `findEntryLocation` answers "is this plugin configured anywhere?", but the bar-widget branch needs "is this widget on the bar?" — and `inBar()` already answers exactly that.

`putBarWidget` guards with `inBar(id)` correctly, then delegates to `setEnabled` and inherits the same conflation:

https://github.com/omacom/omarchy/blob/quattro/shell/services/PluginRegistry.qml#L321-L339

Its `""` return means three different things — "already on the bar", "not mine to touch", and "placed it" — so `omarchy-bar`'s "is on the bar" message can't distinguish the no-op from the success it was written for. The comment at `bin/omarchy-bar:300` says both outcomes are the same answer to the caller; this is a third outcome wearing the same answer.

## Suggested fix

In the bar-widget branch of `setEnabled`, test bar presence rather than configuration presence — `findBarLocation` / `inBar` instead of `findEntryLocation` — so a widget listed only in `plugins[]` still gets placed. A widget in both places is arguably worth cleaning up out of `plugins[]` at the same time.

Happy to send a PR if the direction looks right.

---
Found while debugging a bar-widget that hosts other widgets. Reported with the help of Claude Code.