feat: add plugin system with grpc runtime and native ui plugins (#1742)

* feat: add plugin system with grpc runtime and native ui plugins

- go plugin backends ship as release tarballs (per-platform binaries, checksums) that
  kandev installs from url/upload or filesystem sideload+sync, then spawns and
  supervises via hashicorp/go-plugin (unix socket / loopback tcp+automtls)
- strict typed grpc contract (kandev.plugin.v1): DeliverEvent/InvokeTool/HandleWebhook
  plus a capability-gated host service (state, secrets, emit); author sdk in
  pkg/pluginsdk; packaging cli in cmd/plugin-pack
- native frontend bundles: window.registerKandevPlugin + registry/host loader with
  dynamic routes, nav items, slots and ws handlers; plugins settings page with
  install dialog, sync button and enable/disable/uninstall
- gated behind the plugins feature flag (prod off, dev/e2e on); covered by unit and
  integration tests plus playwright e2e against a real packaged fixture plugin

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): hydrate base app context on plugin routes

Plugin-owned SPA routes classify as RouteUnknown, so the boot payload shipped
only feature flags — no workspaces, workflows, or kanban context. Native plugin
UI (e.g. host.ui.TaskCreateDialog) then had an empty store. Hydrate the same
base context as the Home route for unknown SPA routes.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* feat(plugins): expose TaskCreateDialog on host.ui

Lets a native plugin open kandev's real create-task modal (prefilled via
initialValues) instead of POSTing directly.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): harden filesystem/URL sinks against CodeQL findings

Add sink-level guards on the plugin trust boundary: securejoin containment on
archive extraction (zip-slip), plugin-id validation before every fs_store path,
http/https scheme validation on install-from-URL, and non-interpolated ws-handler
error logging.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): address review findings (races, resource caps, lifecycle)

- serialize per-plugin startup to close a TOCTOU double-spawn race, and let a
  plugin restart after exhausting attempts
- bound plugin startup with an explicit 30s timeout
- reject backslash paths and cap decompressed bytes by actual size in pkgtar;
  only mark a package Signed when a verifier actually ran and passed
- delete plugin_state on uninstall; validate webhook keys against the manifest
  and cap webhook body size; restart an active plugin when a new version installs

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): sanitize plugin-id path sinks for CodeQL

Pin the plugin id to a single path segment with filepath.Base at recordPath/
configPath and pkgtar.Remove — a barrier the taint analysis recognizes.
Behaviour-identical for valid ids (already validated by safePluginID); clears
the go/path-injection alerts.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): inline plugin-id validation at path sinks for CodeQL

filepath.Base does not strip '..', so it is not a path-injection barrier. Call
the recognized safePluginID / id-pattern guard inline in Delete/GetConfig/
SetConfig and pkgtar.Remove — the same barrier that already cleared the Get sink.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): address code-review findings (backend)

- add capability-gating tests for the Host RPC service (was untested)
- close a stop-during-start orphan race so Disable/Uninstall can't leave a
  supervised subprocess behind
- on failed upgrade, remove only the new version dir (not all versions + data);
  restart the prior version
- validate manifest version and reject ids ending in .config (record-file alias)
- make Deliverer.Refresh non-blocking and cancel in-flight delivery on worker stop
- clamp out-of-range webhook status, skip-and-warn corrupt records in List,
  write records atomically, semver-aware sideload version pick

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): address code-review findings (frontend)

- wrap plugin routes + settings routes in a shared PluginErrorBoundary so a
  throwing plugin page shows a fallback instead of white-screening the SPA
- fix disable/re-enable in one session: keep the cached registration and
  re-run initialize instead of relying on a cached ES-module re-import
- prefix root-relative bundle/style URLs with apiBaseUrl (split-origin/Tauri)
- surface partial-install warnings via toast.warning instead of a bare success
- dedupe/remove injected style links on unload; gate PluginNavItems on the flag

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): serialize per-plugin lifecycle + guard config writes

- add a per-plugin keyed lock in Service so concurrent Enable/Disable/Install/
  Uninstall/UpdateConfig for one id are serialized (no status-machine race)
- take s.mu in SetConfig/GetConfig and write config atomically
- bound each plugin initialize() with a timeout so one hung bundle can't stall
  the rest of the boot load

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* test(plugins): split service_test.go to satisfy file-length lint

Move the install/URL test group into service_install_test.go; no logic change.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* ci: build and ship the e2e fixture plugin package to E2E shards

The plugin E2E specs' global-setup requires apps/backend/.build/kandev-plugin-e2e-1.0.0.tar.gz.
Build it in the E2E build job, upload it as an artifact, and download it in both
the regular and container shard jobs.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

---------

Co-authored-by: Kandev Agent <agent@kandev.dev>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
João Salavisa
2026-07-17 12:00:26 +01:00
committed by GitHub
parent c633b8faa1
commit e145d48f22
182 changed files with 22841 additions and 377 deletions
+124
View File
@@ -0,0 +1,124 @@
# Kandev Plugin API contract (native JS UI plugins — "option C")
This is the frozen interface every frontend + example task builds against. Do not
diverge without updating this file.
## Loading model
1. Backend boot payload gains `plugins: ActivePlugin[]` where
`ActivePlugin = { id: string; name: string; bundleUrl: string; styleUrls?: string[] }`.
`bundleUrl` = `/api/plugins/{id}/bundle` — kandev serves this **directly from the
extracted package directory** on local disk
(`~/.kandev/plugins/<id>/<version>/ui/...`, per manifest `ui.bundle`). There is no
reverse proxy and no live upstream request: the plugin subprocess does not need to
be running to serve the UI bundle, since installation already extracted the file.
2. On SPA boot, the **plugin host** (`apps/web/lib/plugins/host.ts`) iterates
`bootPayload.plugins`, injects any `styleUrls` as `<link>`, and dynamically
`import(/* @vite-ignore */ bundleUrl)` each bundle as a native ES module.
3. Each bundle, when evaluated, calls the global:
```ts
window.registerKandevPlugin(pluginId, {
initialize(registry, host): void | Promise<void>,
destroy?(): void,
})
```
4. After the module resolves, the host calls `initialize(registry, host)`. On
plugin disable/uninstall the host calls `destroy?.()` and unregisters everything
that plugin added (registrations are tracked per pluginId).
## Global entry point
`window.registerKandevPlugin(id: string, plugin: KandevPlugin)` — defined by the
host before any bundle loads. Bundles are authored with React as an **external**;
they must use `host.React` (NOT bundle their own React) to share the host instance.
## `host: PluginHostApi`
```ts
interface PluginHostApi {
pluginId: string;
React: typeof import("react"); // host React instance (shared)
jsx: typeof React.createElement; // convenience alias (h)
store: { // kandev app store (zustand StoreApi)
getState(): AppState;
setState(partial): void;
subscribe(listener): () => void;
};
api: {
// fetch scoped to this plugin's backend via kandev proxy:
// GET/POST {method} /api/plugins/{id}/... ; returns parsed JSON
fetch(path: string, init?: RequestInit): Promise<Response>;
};
ui: Record<string, unknown>; // curated @kandev/ui components (Button, Card, Badge...)
theme: "light" | "dark";
}
```
## `registry: PluginRegistry`
```ts
interface NavItem { id: string; label: string; path: string; icon?: string; section?: "main" | "settings"; }
interface PluginRegistry {
// Top-level SPA route, e.g. "/jira". Component rendered by the SPA route resolver
// when window.location path === path (exact match; trailing segments via ":param" not
// required for v1 — exact + startsWith("/plugins/{id}") allowed).
registerRoute(path: string, Component: React.ComponentType): void;
// Sidebar/main nav entry. Rendered by <PluginNavItems/> in the app sidebar.
registerNavItem(item: NavItem): void;
// Route under /settings/plugins/{id}/... rendered inside settings shell.
registerSettingsRoute(path: string, Component: React.ComponentType): void;
// Named slot injection. Host renders all components registered for a slot via
// <PluginSlot name="..." props={...}/>. Initial slots: "task-sidebar",
// "settings-nav", "main-nav-footer".
registerComponent(slot: string, Component: React.ComponentType<{ slotProps?: unknown }>): void;
// WS action handler. Bridged into the existing lib/ws dispatch; called with the
// decoded message payload for that action string.
registerWsHandler(action: string, handler: (payload: unknown) => void): void;
}
```
## Registry internals (host side)
`apps/web/lib/plugins/registry.ts` holds a singleton `PluginRegistry` whose data
is reactive (a small zustand store or event emitter) so host React components
re-render when registrations change. Every registration records the owning
`pluginId` so the host can bulk-unregister on disable. Exposes read selectors:
`getRoutes()`, `getNavItems()`, `getSettingsRoutes()`, `getSlotComponents(slot)`,
`getWsHandlers(action)`.
## Integration points the app must add (task-19)
- `src/spa-routes.tsx`: after the static route switch, before the not-found
fallback, consult `registry.getRoutes()` for a matching path and render it inside
the normal app shell.
- `src/settings-routes.tsx`: consult `registry.getSettingsRoutes()` for
`/settings/plugins/{id}/*` paths.
- App sidebar (grep the nav list component): render `<PluginNavItems/>` reading
`registry.getNavItems()`.
- `lib/ws/router.ts` / `lib/ws/client.ts`: after built-in dispatch, forward the
message to any `registry.getWsHandlers(action)`.
- `components/plugins/plugin-slot.tsx`: `<PluginSlot name props/>` renders all
slot components; drop into task detail sidebar + settings nav as initial hosts.
## Security posture (documented, enforced where cheap)
Plugin JS runs in the kandev origin with store access — this is the accepted
tradeoff of option C. v1 mitigations: only **active, operator-installed** plugins
load; bundles are served by kandev from the extracted package directory (same-origin,
no third-party CDN, no upstream network hop); host wraps `initialize` in try/catch so
a broken plugin can't crash boot; registrations namespaced + bulk-revocable per
plugin. No credentials are ever displayed to the operator — installing a plugin (via
URL or upload) has nothing to copy or reveal, unlike the old register flow's one-time
API key/webhook secret. Sandboxing plugin JS (worker/realms) is explicit future work.
## Example plugin must (task-21)
Ship a bundle that on `initialize` registers: a nav item "Hello" → route
`/plugins/hello` rendering a native page (uses `host.jsx` + `host.ui`), a
`task-sidebar` slot component, and a WS handler for `task.created` that updates a
counter in its page via the plugin's own module state. No bundled React.