pastilhasandClaude Opus 5
631443e5ae
naming officerdev/chat in layout.ts rendered a second Hosts panel. the shared
registry is not the thing that decides: PluginScreen renders the workspace with
appTypes={{ allowed, fallback: allowed[0] }} where allowed is THIS plugin's own
panel keys, so an unrecognised appType silently becomes the first panel. that
boundary is deliberate — its comment says a persisted layout must not be able
to render another plugin's panel inside this screen — and it is a different
mechanism from seedAppRegistry merging the panels, which is what the last
commit's reasoning confused it with.
so the panel is ours and EmbeddableChat inside it is the platform's. it is
exported from officerdev for exactly this: the whole composer, transcript,
queue and model picker, with no second implementation to fall behind the
first.
it carries a two-sentence promptPrefix, which is per-panel and costs nobody
else's conversation anything. deliberately short: the skill holds the
knowledge and reaches every session for two lines, so anything longer here
would be the skill written twice and this would be the stale copy. all it says
is where the agent is standing, which is the one thing the skill cannot know.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Officer-Session: 656f5c85-e49a-4f1f-b9f3-042a04b4d043
39 lines
2.0 KiB
TypeScript
39 lines
2.0 KiB
TypeScript
import { EmbeddableChat } from 'officerdev';
|
||
|
||
/**
|
||
* A chat, sitting under the two lists.
|
||
*
|
||
* ── Why this exists instead of naming `officerdev/chat` in the layout ──
|
||
*
|
||
* It was tried, and it produced two Hosts panels. `PluginScreen` renders the workspace with
|
||
* `appTypes={{ allowed, fallback: allowed[0] }}` where `allowed` is **this plugin's own panel keys**, so a
|
||
* layout naming anything else silently falls back to the first one. That is deliberate and its comment
|
||
* says why: a persisted layout must not be able to render another plugin's panel inside this screen.
|
||
*
|
||
* The registry being shared is a separate thing from the layout being allowed to name what is in it. Both
|
||
* are true; only the second one decides this.
|
||
*
|
||
* So the panel is ours and the chat inside it is the platform's. `EmbeddableChat` is exported from
|
||
* `officerdev` for precisely this — the whole composer, transcript, queue and model picker, with no second
|
||
* implementation to fall behind the first.
|
||
*
|
||
* ── The prefix ──
|
||
*
|
||
* `promptPrefix` is described in `ChatPanelWrapper` as "a system prompt only this app understands". It is
|
||
* per-panel rather than per-session-wide, so it costs nothing to anybody else's conversation.
|
||
*
|
||
* Kept to two sentences on purpose. The Proxy Manager SKILL carries the real knowledge and reaches every
|
||
* conversation on the machine at a cost of two lines; this only says where the agent is standing, which is
|
||
* the one thing the skill cannot know. Anything longer would be the skill written twice, and the copy here
|
||
* would be the stale one.
|
||
*/
|
||
const PREFIX = [
|
||
'This chat is a panel on the Proxy Manager screen, beside a list of this machine’s proxy hosts and a list',
|
||
'of its TLS certificates. Questions about domains, forwarding, whether a site is up, or when a certificate',
|
||
'expires are about THIS machine — read the Proxy Manager skill before answering one.',
|
||
].join(' ');
|
||
|
||
export const ChatPanel = ({ panelId }: { panelId: string }) => (
|
||
<EmbeddableChat className="min-h-0 flex-1" panelId={panelId} promptPrefix={PREFIX} />
|
||
);
|