test(agent): pin the agent row actions - AgentActions (Closes #1836)

[dmitrii-f-t27]

Closes #1836

What

The first suite for apps/agent/entrypoints/app/agents/agent-row/AgentActions.tsx (AgentActions, the only export), under bun test in the agent group.

• describe block: AgentActionsTsxContract (appears nowhere else in the tree).
• Minimal mock surface: only the router hook is stubbed (query-qualified spread-real read keeping every other export); the dropdown and tooltip kits stay real (their triggers survive a static render, as the open-in-chat and web-preview suites show), and the display helpers (displayName, canDelete, canRename) stay real — the aria-label pin exercises the real displayName against the fixture agent.
• Pin: an active turn draws the filled Resume button and no active turn the outlined Chat one; both name the agent on the more-actions trigger (More actions for Research Bot); the dropdown's menu content stays hidden as the menu first mounts. The menu items, coming-soon tooltips and the delete/copy/navigate flows sit behind the kits' internal open states and a live document — named in the suite header per User Story 2.
• FR-004 shown both ways, quoted: broken subject (Resume button relabelled) → 0 pass / 1 fail; restored → 1 pass / 0 fail / 8 expect(). git diff --name-only lists only the new test file.

Local gates

• Suite: 1 pass / 0 fail (8 expect()); grep -c 'it(' = 1; 1 + 0 blocked = 1.
• Group (apps/agent): 262 pass / 3 fail / 265 tests — the 3 = the pre-existing port-helper noise, unchanged from baseline.
• Biome: 0 (no fixes needed). Typecheck: 19 = base, 0 mentions of the new file.

No database, no container, no network.

[tri-claw]

Summary by t27-bees

This PR adds a contract test for the AgentActions component to pin its current rendering behavior (Resume vs Chat button based on activeTurnId, more-actions trigger label, dropdown menu hidden on mount) so future changes have a baseline to fail against. Only the router's useNavigate is stubbed; the dropdown, tooltip kits, and display helpers remain real.

Worth a look

• AgentActions.test.ts:28 – useNavigate is mocked as () => () => {}, but react-router's useNavigate returns a function accepting (to: To, options?: NavigateOptions). If the component ever calls navigate(path) during render (e.g., in an effect), the mock swallows the call silently; a stricter mock or vi.fn() would surface misuse.
• AgentActions.test.ts:15–18 – The query-qualified imports (react-router?agent-actions-real, ./AgentActions?agent-actions-suite-real) rely on Bun's module re-instantiation feature. This is non-standard and may break if Bun changes its import reflection semantics or if the test runs under a different runtime.
• AgentActions.test.ts:55–65 – Assertions check for raw strings ('Resume', 'Chat', 'More actions for Research Bot'). These are coupled to display copy; any i18n, redesign, or displayName change will fail the test without signaling a real regression.
• AgentActions.test.ts:48 – The fixture hard-codes canChat: true, canDelete: true. The test doesn't cover the canChat=false or canDelete=false branches, so a prop-driven render difference would go unnoticed.
• AgentActions.test.ts:50 – renderToStaticMarkup produces a static HTML string; it does not execute effects, ref callbacks, or event handlers. The dropdown's open-state content (Copy id, Delete) is correctly hidden on mount, but the test cannot verify that clicking the trigger reveals them—a gap called out in the header but still a blind spot.

---

_{Read: 1 files, +92 -0, the whole diff. Head 1c3ab9c. 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 — 3637/3707

Suite	Passed	Failed	Skipped
✅ agent	265/265	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