From ac3d7ddae7feb60771d1ad3ce7c3126b8d0f2bea Mon Sep 17 00:00:00 2001 From: Teknium <127238744+teknium1@users.noreply.github.com> Date: Wed, 19 Aug 2026 00:39:10 -0700 Subject: [PATCH] fix(desktop): skill hub installs on non-default profiles no longer 404, and failed installs surface MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Three fixes for the "Install on this agent" pipeline, covering the whole split-brain class between action-spawning endpoints and their status polls: 1. electron/connection-config.ts — the /api/actions/{name}/status poll family now routes to the same backend as every action-spawning route. Before, POST /api/skills/hub/install ran on the PRIMARY backend (scoped route) while the follow-up status poll for a non-default profile routed to the profile's POOLED backend, which never registered the dynamic action name (skills-install-- lives only in the spawning process's memory) -> 404 "Unknown action" toast even though the install succeeded. POST /api/mcp/catalog/install joins the scoped table for the same reason. 2. src/store/hub-actions.ts — a non-zero subprocess exit now rejects with the action log tail so the caller's catch toasts it. Before, a failed install (scan gate, network, bad identifier) stopped silently: no toast, no row flip, and the unchanged skills list read as "install did nothing". 3. src/contrib/runtime-loader.ts — a disk plugin copy shadowed by a bundled twin now publishes a visible "(stale disk copy)" inventory row carrying the folder path, instead of a console.info nobody sees. Stale desktop-plugins/ leftovers from dev deploys are the same folders that actively break the feature on shells without the bundled twin. --- .../electron/connection-config.test.ts | 31 +++++++ apps/desktop/electron/connection-config.ts | 16 +++- .../src/contrib/runtime-loader.test.ts | 55 ++++++++++- apps/desktop/src/contrib/runtime-loader.ts | 13 ++- apps/desktop/src/store/hub-actions.test.ts | 91 +++++++++++++++++++ apps/desktop/src/store/hub-actions.ts | 11 +++ 6 files changed, 213 insertions(+), 4 deletions(-) create mode 100644 apps/desktop/src/store/hub-actions.test.ts diff --git a/apps/desktop/electron/connection-config.test.ts b/apps/desktop/electron/connection-config.test.ts index 5bce688fc2..8bc40e639a 100644 --- a/apps/desktop/electron/connection-config.test.ts +++ b/apps/desktop/electron/connection-config.test.ts @@ -718,6 +718,37 @@ test('resolveProfileApiRequest scopes complete safe families according to their ) }) +test('resolveProfileApiRequest routes action-status polls with the action-spawning routes', () => { + // /api/actions/{name}/status must land on the SAME backend as the endpoints + // that spawn actions (skills hub install/uninstall/update, mcp catalog + // install): _spawn_hermes_action registers the dynamic action name only in + // the spawning process. Splitting the pair 404s the poll with + // "Unknown action: skills-install--". + assert.deepEqual( + resolveProfileApiRequest('iris', '/api/actions/skills-install-ascii-art-dd7bccf1/status?lines=200', { + requestMethod: 'GET' + }), + { + backendProfile: null, + requestPath: '/api/actions/skills-install-ascii-art-dd7bccf1/status?lines=200&profile=iris' + } + ) + // The spawn side (hub install) and the poll side must agree on the backend. + assert.deepEqual( + resolveProfileApiRequest('iris', '/api/skills/hub/install', { + requestMethod: 'POST' + }), + { backendProfile: null, requestPath: '/api/skills/hub/install?profile=iris' } + ) + // MCP catalog installs spawn background actions too — same pairing rule. + assert.deepEqual( + resolveProfileApiRequest('iris', '/api/mcp/catalog/install', { + requestMethod: 'POST' + }), + { backendProfile: null, requestPath: '/api/mcp/catalog/install?profile=iris' } + ) +}) + test('resolveProfileApiRequest preserves remote routing precedence', () => { assert.deepEqual( resolveProfileApiRequest('iris', '/api/memory/reset', { diff --git a/apps/desktop/electron/connection-config.ts b/apps/desktop/electron/connection-config.ts index 74b5aace58..1934fbec0d 100644 --- a/apps/desktop/electron/connection-config.ts +++ b/apps/desktop/electron/connection-config.ts @@ -548,7 +548,11 @@ const LOCAL_PRIMARY_SCOPED_ROUTES = new Set([ 'GET /api/skills/hub/search', 'GET /api/skills/hub/sources', 'POST /api/skills/hub/uninstall', - 'POST /api/skills/hub/update' + 'POST /api/skills/hub/update', + // Spawns a background action polled via /api/actions/{name}/status — must + // live on the SAME backend as that poll family (below), or the poll asks a + // backend that never registered the dynamic action name and 404s. + 'POST /api/mcp/catalog/install' ]) function localPrimaryRequestScope(opts: ProfileRouteOptions): boolean | null { @@ -572,6 +576,16 @@ function localPrimaryRequestScope(opts: ProfileRouteOptions): boolean | null { return true } + // Action-status polls MUST land on the same backend as the endpoints that + // spawned them: `_spawn_hermes_action` registers the (often dynamic, e.g. + // `skills-install--`) action name only in the spawning + // process's memory. Every action-spawning route above scopes to the + // primary, so the poll family follows — a pooled-backend poll 404s with + // "Unknown action" even though the install itself succeeded (#89xxx). + if (pathname.startsWith('/api/actions/')) { + return true + } + // Every current /api/tools handler accepts `profile`; every /api/profiles // handler either aggregates profiles or names its target in the path/body. // These are the only whole families safe to route through the primary. diff --git a/apps/desktop/src/contrib/runtime-loader.test.ts b/apps/desktop/src/contrib/runtime-loader.test.ts index 17a0661c24..0ed142e17f 100644 --- a/apps/desktop/src/contrib/runtime-loader.test.ts +++ b/apps/desktop/src/contrib/runtime-loader.test.ts @@ -3,8 +3,8 @@ import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest' import type { HermesReadDirResult } from '@/global' import type * as HermesModule from '@/hermes' -import { $pluginRecords, setPluginEnabled } from './plugins-store' -import { discoverRuntimePlugins, watchRuntimePlugins } from './runtime-loader' +import { $pluginRecords, publishPlugin, setPluginEnabled } from './plugins-store' +import { discoverRuntimePlugins, loadRuntimePlugin, watchRuntimePlugins } from './runtime-loader' // getStatus would supply the connected backend's hermes_home — a REMOTE path in // remote mode. The disk scanner must NOT derive the plugin root from it (#66899). @@ -178,3 +178,54 @@ describe('watchRuntimePlugins dir watch (#66899)', () => { expect(getStatus).not.toHaveBeenCalled() }) }) + +describe('bundled-shadowed disk copies', () => { + it('skips a disk copy of a bundled plugin but publishes a visible inventory row', async () => { + // The bundled twin is already registered (build-time glob). + publishPlugin({ id: 'hermes-bots', name: 'Bot Mode', kind: 'bundled', status: 'loaded' }) + + // Same blob→data: URL reroute as the opt-in test above. + const createObjectURL = vi + .spyOn(URL, 'createObjectURL') + .mockImplementation( + blob => + `data:text/javascript;base64,${Buffer.from((blob as unknown as { parts: string[] }).parts.join('')).toString('base64')}` + ) + + const revokeObjectURL = vi.spyOn(URL, 'revokeObjectURL').mockImplementation(() => undefined) + const RealBlob = globalThis.Blob + vi.stubGlobal( + 'Blob', + class { + parts: string[] + constructor(parts: string[]) { + this.parts = parts + } + } + ) + + try { + const id = await loadRuntimePlugin( + 'export default { id: "hermes-bots", name: "Bot Mode", register() {} }', + 'hermes-bots', + { file: '/local/.hermes/desktop-plugins/hermes-bots/plugin.js' } + ) + + // Skipped — the bundled copy stays the only live registration... + expect(id).toBeNull() + expect($pluginRecords.get()['hermes-bots']).toMatchObject({ kind: 'bundled', status: 'loaded' }) + + // ...but the stale folder is DISCOVERABLE: an inventory row names it, + // carries its path (reveal/delete affordance), and can never activate. + expect($pluginRecords.get()['hermes-bots:disk-shadowed']).toMatchObject({ + kind: 'disk', + status: 'disabled', + file: '/local/.hermes/desktop-plugins/hermes-bots/plugin.js' + }) + } finally { + createObjectURL.mockRestore() + revokeObjectURL.mockRestore() + vi.stubGlobal('Blob', RealBlob) + } + }) +}) diff --git a/apps/desktop/src/contrib/runtime-loader.ts b/apps/desktop/src/contrib/runtime-loader.ts index c99f0de6e4..ba4c7fd0c4 100644 --- a/apps/desktop/src/contrib/runtime-loader.ts +++ b/apps/desktop/src/contrib/runtime-loader.ts @@ -146,9 +146,20 @@ export async function loadRuntimePlugin( // standalone install of hermes-bots predating its adoption in-tree) must // not register a second time: contributions would double up and the two // copies would fight over storage. The bundled copy wins; the disk copy - // is skipped quietly so stale installs keep working after an app update. + // is skipped — but VISIBLY: a silent skip left the stale folder + // undiscoverable while (on shells without the bundled twin) the same + // folder actively breaks the feature it shadows. The inventory row + // carries the file path so Settings → Plugins can reveal it for deletion. if ($pluginRecords.get()[plugin.id]?.kind === 'bundled') { console.info(`[plugins] ${origin} skipped — "${plugin.id}" already ships bundled with the app`) + publishPlugin({ + id: `${plugin.id}:disk-shadowed`, + name: `${plugin.name ?? plugin.id} (stale disk copy)`, + description: `Shadowed by the bundled "${plugin.id}" plugin — this folder is no longer used and can be deleted.`, + kind: options.kind ?? 'disk', + file: options.file, + status: 'disabled' + }) return null } diff --git a/apps/desktop/src/store/hub-actions.test.ts b/apps/desktop/src/store/hub-actions.test.ts new file mode 100644 index 0000000000..6513e5bac8 --- /dev/null +++ b/apps/desktop/src/store/hub-actions.test.ts @@ -0,0 +1,91 @@ +import { beforeEach, describe, expect, it, vi } from 'vitest' + +// A failed hub action (non-zero subprocess exit) must REJECT so callers toast +// it — the Aug 2026 report shape was a scan-gated/failed `hermes skills +// install` exiting non-zero with no toast, no list change, and no error +// anywhere in the UI: the user read it as "install did nothing". + +const getActionStatus = vi.fn() +const installSkillFromHub = vi.fn() + +vi.mock('@/hermes', () => ({ + getActionStatus: (...args: unknown[]) => getActionStatus(...args), + installSkillFromHub: (...args: unknown[]) => installSkillFromHub(...args), + uninstallSkillFromHub: vi.fn(), + updateSkillsFromHub: vi.fn() +})) + +vi.mock('@/lib/query-client', () => ({ + queryClient: { invalidateQueries: vi.fn() } +})) + +vi.mock('@/lib/slash-completion-cache', () => ({ + invalidateSlashCompletions: vi.fn() +})) + +vi.mock('@/store/activity', () => ({ + upsertDesktopActionTask: vi.fn() +})) + +// hub-actions subscribes to the active gateway profile at module scope (its +// per-profile state wipe); a minimal stub keeps this test off the real store. +vi.mock('@/store/profile', () => ({ + $activeGatewayProfile: { subscribe: vi.fn() }, + normalizeProfileKey: (value: unknown) => String(value ?? '') || 'default' +})) + +import { $hubActions, $hubInstalledOverride, installHubSkill } from './hub-actions' + +describe('installHubSkill failure surfacing', () => { + beforeEach(() => { + getActionStatus.mockReset() + installSkillFromHub.mockReset() + $hubActions.set({}) + $hubInstalledOverride.set({}) + }) + + it('rejects with the action log tail when the subprocess exits non-zero', async () => { + installSkillFromHub.mockResolvedValue({ name: 'skills-install-ascii-art-dd7bccf1' }) + getActionStatus.mockResolvedValue({ + name: 'skills-install-ascii-art-dd7bccf1', + running: false, + exit_code: 1, + pid: 4242, + lines: ['Resolving ascii-art...', 'Error: security scan blocked install'] + }) + + await expect(installHubSkill('ascii-art')).rejects.toThrow('security scan blocked install') + + // A failed install must not paint the row as installed. + expect($hubInstalledOverride.get()['ascii-art']).toBeUndefined() + // ...but the row's running flag still settles to false for the button. + expect($hubActions.get()['ascii-art']?.running).toBe(false) + }) + + it('rejects with the exit code when the log carries no detail', async () => { + installSkillFromHub.mockResolvedValue({ name: 'skills-install-x-00000000' }) + getActionStatus.mockResolvedValue({ + name: 'skills-install-x-00000000', + running: false, + exit_code: 7, + pid: 4242, + lines: [] + }) + + await expect(installHubSkill('x')).rejects.toThrow('exited with code 7') + }) + + it('resolves and flips the row on a clean exit', async () => { + installSkillFromHub.mockResolvedValue({ name: 'skills-install-ok-11111111' }) + getActionStatus.mockResolvedValue({ + name: 'skills-install-ok-11111111', + running: false, + exit_code: 0, + pid: 4242, + lines: ['Installed ok'] + }) + + await expect(installHubSkill('ok')).resolves.toBeUndefined() + expect($hubInstalledOverride.get()['ok']).toBe(true) + }) +}) diff --git a/apps/desktop/src/store/hub-actions.ts b/apps/desktop/src/store/hub-actions.ts index 1f0bf30bd3..23e5382310 100644 --- a/apps/desktop/src/store/hub-actions.ts +++ b/apps/desktop/src/store/hub-actions.ts @@ -122,6 +122,17 @@ async function runHubAction( // …and the composer's `/` list, which caches the command catalog for an // hour and would otherwise keep offering the skill we just removed. invalidateSlashCompletions() + + // A non-zero exit is a real failure — throw so the caller's catch toasts + // it. Before this, a failed subprocess (scan gate, network, bad + // identifier) just stopped silently: no flip, no toast, and the user read + // the unchanged skills list as "install did nothing" (Aug 2026 report). + // The last log lines carry the subprocess's actual error. + if (exitCode !== null && exitCode !== 0) { + const detail = ($hubActions.get()[key]?.lines ?? []).slice(-3).join('\n').trim() + + throw new Error(detail || `Action exited with code ${exitCode}`) + } } catch (err) { // A profile switch points the next poll at the new backend, which 404s the // old action name — that's an abandonment, not a failure, so swallow it