feat: mount discovered module UI in a minimal Ekklesia browser host
CI / verify (pull_request) Successful in 34s
CI / infrastructure-smoke (pull_request) Successful in 21s
CI / migration-smoke (pull_request) Successful in 43s

Phase 6C pass 2. Every link in the module UI chain existed except the one
that closes it: a module could be authored, built, validated, published at a
browser URL, and loaded by a generic loader — and nothing in Ekklesia did any
of it. There was no browser application at all.

Two pieces were missing. A browser could not discover what to mount, and it
must not work that out for itself.

GET /api/v1/module-ui is the inventory: per publishable module, its id and
version, the server-produced artifact URL, the server-produced root of the
module's own API, and the subset of that module's declared permissions the
caller holds. Five fields, each because building a ModuleUiHostContext needs
it — no filesystem paths, and no name/icon/folder/order, because navigation
metadata is a design that does not exist yet. Its module set is the ADR 0025
publication registry rather than a rescan, so the inventory and the artifact
route cannot disagree. It requires a new Core permission, core.modules.ui.read
(baseline authenticated), because it is precisely the enumeration the anonymous
artifact route deliberately withholds.

Both URLs are server-produced. moduleUiArtifactUrl already existed;
moduleApiBaseUrl is now exported and moduleRoutePath builds on it, so each URL
shape has exactly one author. Without it the shell would hardcode
/api/v1/modules/<id> — the prefix ADR 0022 forbids a module UI from spelling,
reappearing in the host instead.

surfaces/shell (@ekklesia/shell) is the repository's first surface: a small
React application built to static files by its own esbuild script. It resolves
the principal, fetches the inventory, builds the host context entirely from
server-derived data, and hands the artifact URL to @ekklesia/module-ui-host.
One frame per selection — keyed by module identity — defeats the stale-load
race; the host owns the container, the loading state, and every failure, so a
broken module costs its own panel rather than the shell.

The shell statically imports no module, and both levels are checked: no source
import may name a module, and the emitted bundle must contain exactly one
dynamic import (the loader's) with no module id, no /api/v1/modules, no eval,
and no blob URL. Permission hints come from Core's own evaluator and stay
presentation-only — the end-to-end test has a UI holding no grant call its API
anyway and still receive 403.

Surfaces may now depend only on browser-safe shared roles; @ekklesia/contracts
is deliberately excluded, since its barrel pulls Drizzle's types and Node's
globals into the browser — the hole ADR 0024 closed one layer down.

Not the Dashboard: no navigation, grouping, icons, ordering, design system, or
routing. The selector is a list of module ids.

Co-Authored-By: Claude Opus 5 <[email protected]>
This commit is contained in:
ekklesia
2026-08-22 08:55:44 -05:00
co-authored by Claude Opus 5
parent 7a76ff6f26
commit 616c1d4834
46 changed files with 3629 additions and 36 deletions
@@ -0,0 +1,115 @@
import { loadModuleUi, type ModuleUiImporter } from "@ekklesia/module-ui-host";
import type { ModuleUiPrincipal } from "@ekklesia/module-ui-sdk";
import { type ReactElement, useEffect, useRef, useState } from "react";
import type { EkklesiaHostConfig, HostModuleUiDescriptor } from "../host-protocol.js";
import { createModuleUiContext, moduleUiSourceUrl } from "../module-context.js";
type FrameStatus = "loading" | "ready" | "failed";
export interface ModuleUiFrameProps {
readonly config: EkklesiaHostConfig;
readonly descriptor: HostModuleUiDescriptor;
readonly principal: ModuleUiPrincipal;
/** Test-only seam, forwarded to the generic loader so a test can hold an import pending and prove
* the stale-load rule. Production composition leaves it unset and the loader's own
* `(url) => import(url)` runs. */
readonly importModule?: ModuleUiImporter;
}
/**
* Mounts exactly one module UI into a container this component owns, and tears it down again.
*
* ## Host-owned container, module-owned contents
*
* The `<div>` below belongs to the shell. The module fills it and must return an unmount that
* empties it; it never receives the shell's header, the module list, the page root, or the document.
* The loading and error states are rendered as siblings, so a module can never own the shell's
* pre-mount or failure state — which is the point of the host owning them.
*
* ## One frame per selection, which is what defeats the stale-load race
*
* The parent renders this with a `key` of the module's identity, so selecting a different module
* unmounts this component entirely and creates a fresh one with a **fresh container element**. A
* slow import that resolves after the selection changed therefore mounts into a node that is already
* detached from the document and is immediately unmounted by this component's own cleanup — it can
* never render over, or into, the newer selection's container. The `cancelled` flag makes that
* explicit rather than incidental: a load that finishes after cleanup ran is torn down at once and
* never reaches `setStatus`, so it also cannot resurrect a stale status on a component React has
* already discarded.
*
* ## Failure is the host's to render
*
* Every way this can fail — the artifact will not fetch, the bundle is not a module, its declared
* identity does not match the manifest, or `mount` throws — is caught here and turned into a fixed,
* host-owned message. Nothing from the caught error reaches the DOM: not a URL, not a path, not a
* stack. The error object goes to the browser console, which is where a developer's diagnostics
* belong and where ADR 0024 intends the loader's `cause` to be read. The rest of the shell keeps
* rendering, so a broken module costs its own panel and nothing else.
*/
export function ModuleUiFrame(props: ModuleUiFrameProps): ReactElement {
const containerRef = useRef<HTMLDivElement | null>(null);
const [status, setStatus] = useState<FrameStatus>("loading");
const { config, descriptor, importModule, principal } = props;
useEffect(() => {
const container = containerRef.current;
if (!container) {
return;
}
let cancelled = false;
let mounted: Awaited<ReturnType<typeof loadModuleUi>> | undefined;
setStatus("loading");
void (async () => {
try {
const instance = await loadModuleUi({
container,
context: createModuleUiContext({ config, descriptor, principal }),
...(importModule ? { importModule } : {}),
source: {
moduleId: descriptor.id,
url: moduleUiSourceUrl(config, descriptor),
version: descriptor.version,
},
});
if (cancelled) {
// Selection already moved on. Release it immediately rather than leaving a module running
// in a detached container with no handle to stop it.
await instance.unmount().catch(() => undefined);
return;
}
mounted = instance;
setStatus("ready");
} catch (error) {
console.error("[ekklesia] a module UI failed to load", error);
if (!cancelled) {
setStatus("failed");
}
}
})();
return () => {
cancelled = true;
// Idempotent by contract (ADR 0024), and a throwing cleanup is swallowed here: teardown is
// already in progress, and a module that fails to clean up must not take the shell with it.
void mounted?.unmount().catch(() => undefined);
};
}, [config, descriptor, importModule, principal]);
return (
<section aria-live="polite" style={{ display: "grid", gap: "0.75rem" }}>
{status === "loading" ? (
<p data-testid="module-loading" style={{ color: "#57534e", margin: 0 }}>
Loading module…
</p>
) : undefined}
{status === "failed" ? (
<p data-testid="module-error" style={{ color: "#9f1239", margin: 0 }}>
This module’s interface could not be loaded. Try another module, or ask an administrator
to check the installation.
</p>
) : undefined}
<div data-testid="module-container" ref={containerRef} />
</section>
);
}