test(agent): pin the MCP server settings card - ServerSettingsCard (Closes #1529)

[dmitrii-f-t27]

Closes #1529

What changed

One new file: trios/agent-server/apps/agent/entrypoints/app/mcp-settings/ServerSettingsCard.test.ts (the path the issue's Success Criteria names). The subject ServerSettingsCard.tsx is untouched.

The 1 exported symbol is exercised (1 it( block; 0 blocked). The card reads the BrowserOS adapter prefs, the server messaging and the metrics track, so each seam is stubbed with a full stand-in: the adapter (whose real module cannot initialize outside a browser - the complete stand-in this wave's storage suites run with, every consumer-facing method rejecting the way the real one does outside the shell, as a named stub class), the server messaging module (whose real load pulls the extension polyfill, which refuses outside an extension) answering both exports, and the metrics track (a no-op - its real module pulls the adapter graph other suites stub incompletely). Measured, the static render goes through and shows the card as it first mounts: the MCP Server Settings heading with its description, the external-access switch with its label, and the restart control.

Not pinned, and why (issue User Story 2): the settings flow itself - the remote-access toggle with its confirm dialogs, the server restart wait loop, the pref writes - they live in effects and handlers that need a live document and window. The port env seam is seeded before any import can reach the env module (its one-time snapshot - the trap the agent-outputs suite documented), with the subject imported dynamically for that reason.

How it was verified (FR-004)

• Passing run against the unmodified subject: 1 pass / 0 fail / 4 expect() calls, Ran 1 test across 1 file.
• Deliberately broken copy (the card heading changed, valid syntax): 0 pass / 1 fail / 1 expect() calls. The subject was restored byte-identical (git checkout; git diff --name-only prints nothing) and the suite passes again.
• No network, database or container; no existing test modified or weakened. The file is biome-check clean (an initial noThisInStatic warning on the stub was resolved by extracting the stub to a named class). The local typecheck reports 19 errors - the pre-existing base of this tree - and 0 mentions of the new file. Verified in-process: the full agent group passes this file next to every suite of this wave (256 pass / 3 fail, the 3 pre-existing local-only env failures, green in CI).
• Local reproducibility note (issue bun test cannot resolve @/ aliases once .wxt/tsconfig.json exists (wxt prepare output breaks the bun 1.3.6 runtime resolver) #1909): bun run typecheck runs wxt prepare, whose generated .wxt/tsconfig.json makes bun 1.3.6 stop resolving @/ aliases for test runs until that generated directory is removed. CI's Tests jobs never run wxt prepare; every run quoted above is from a checkout without .wxt/.

Not verified

• Nothing beyond the quoted local runs; the CI Tests / agent job re-runs the suite on this SHA.

Dmitrii Fedorov (@dmitrii-f-t27)

[tri-claw]

Summary by t27-bees

This PR adds a contract test for ServerSettingsCard.tsx that pins its initial static render output. The test stubs three external seams (BrowserOS adapter, server messaging, metrics) and verifies the card shows the heading, description, external-access switch label, and restart button on first mount. Interactive flows (toggle, restart wait loop, pref writes) are intentionally not pinned because they require a live browser environment.

Worth a look

• trios/agent-server/apps/agent/entrypoints/app/mcp-settings/ServerSettingsCard.test.ts:15 — process.env.VITE_BROWSEROS_SERVER_PORT is set before the dynamic import to beat the env module's one-time snapshot. If the env module is imported elsewhere in the test setup (e.g., via a shared test helper), this seeding could be too late and the port would fall back to a different value.

• trios/agent-server/apps/agent/entrypoints/app/mcp-settings/ServerSettingsCard.test.ts:39-41 — The serverMessages mock returns sendServerMessage: async () => ({ healthy: true }). If the component calls sendServerMessage during render (not just in effects), the resolved value could affect the static markup. The test doesn't assert anything about the health state, so a change in that behavior would go unnoticed.

• trios/agent-server/apps/agent/entrypoints/app/mcp-settings/ServerSettingsCard.test.ts:53-60 — The test asserts only the final mounted text content. The suite comment mentions "the loading state answers the skeleton spinner" but no assertion checks for a skeleton/loading UI. If the component initially renders a skeleton that is replaced after data loads, the test would pass even if the skeleton disappeared or changed.

---

_{Read: 1 files, +95 -0, the whole diff. Head 48f1927. Model: openai-compatible/nvidia/nemotron-3-ultra-550b-a55b. Commands: @t27-bees review | summary | help | pause | resume. Free for every repository.}

[github-actions]

✅ Tests passed — 3631/3701

Suite	Passed	Failed	Skipped
✅ agent	259/259	0	0
✅ build	9/9	0	0
✅ cdp-protocol	5/5	0	0
✅ eval	93/93	0	0
✅ server-agent	280/280	0	0
✅ server-api	2159/2228	0	69
✅ server-browser	6/6	0	0
✅ server-integration	10/11	0	1
✅ server-lib	279/279	0	0
✅ server-pglive	166/166	0	0
✅ server-root	68/68	0	0
✅ server-sim	8/8	0	0
✅ server-skills	31/31	0	0
✅ server-tools	244/244	0	0
✅ shared	14/14	0	0

View workflow run

[dmitrii-f-t27]

CI gates on this SHA:

• Tests / agent: pass — the settings-card pin runs in the agent group.
• All other Tests/ matrix jobs pass*, including server-pglive (no flake rerun needed).
• Typecheck job: 19 errors = the known base count; 0 mentions of ServerSettingsCard.test.ts in the job log.
• Biome job: red from base findings; 0 mentions of ServerSettingsCard.test.ts in the job log.

Dmitrii Fedorov (@dmitrii-f-t27)