test: favor behavior over implementation details

[t3-code]

what changed

• removed the unused mobile terminal subscription-key helper and its six implementation-detail tests
• removed the test-only pending-launch peek method, keeping coverage through the production take behavior
• removed a test that inspected a private MessagesTimeline class string, then made the class constant private
• removed the test-only composer primary-action breakpoint alias and kept coverage on the public layout behavior
• removed a duplicate literal assertion from the shared terminal id allocator test
• dropped the overlapping mobile terminal test removal because PR test: remove redundant and stale tests #6267 already owns it

why

these tests and exports exposed private implementation details without adding unique regression signal. the remaining tests exercise pending-launch consumption, rendered timeline behavior, composer layout behavior at the actual breakpoint, and the terminal allocator's public result.

verification

• focused tests: 5 files passed, 52 tests passed
• web typecheck passed
• mobile typecheck passed
• targeted lint passed
• targeted format passed
• git diff --check passed

created by gpt-5.6-sol via hermes agent using openai-codex.

request provenance

• requested by: @juliusmarminge

---

Note

Low Risk
Test and API-surface cleanup only; no production logic changes beyond making one CSS constant non-exported and removing an unused peek helper.

Overview
Trims tests and exports that only locked in private helpers, without changing user-facing behavior.

Mobile terminal: Drops peekPendingTerminalLaunch and pending-launch tests now assert consume-once via repeated takePendingTerminalLaunch. Removes threadTerminalSubscriptionKey and deletes its dedicated test file; attach input building stays.

Web chat: Stops exporting toolCallExpandedBodyClassName from MessagesTimeline (module-local only) and removes the test that asserted on that class string.

Composer layout: Removes the COMPOSER_PRIMARY_ACTIONS_COMPACT_BREAKPOINT_PX alias; primary-action compactness still uses the wide footer breakpoint in tests and implementation.

Shared: Drops a redundant duplicate expectation in nextTerminalId tests.

^{Reviewed by Cursor Bugbot for commit f3359b9. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Remove internal implementation details from exported module APIs in tests

• Removes peekPendingTerminalLaunch from terminalLaunchContext.ts and threadTerminalSubscriptionKey from threadTerminalPanelModel.ts; tests now only assert on observable behavior via takePendingTerminalLaunch.
• Makes toolCallExpandedBodyClassName in MessagesTimeline.tsx and COMPOSER_PRIMARY_ACTIONS_COMPACT_BREAKPOINT_PX in composerFooterLayout.ts internal, removing their exports and corresponding test assertions.
• Tests are updated to verify behavior through public-facing functions rather than inspecting internal constants or helper utilities.

^{Macroscope summarized f3359b9.}

[macroscopeapp]

Approvability

Verdict: Approved f3359b9

Test refactoring that removes implementation-detail tests and makes internal symbols private. No runtime behavior changes - purely removing redundant exports and test code.

^{You can customize Macroscope's approvability policy. Learn more.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	12.6 KiB	12.6 KiB	−11 B (−0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.3 KiB	6.3 KiB	−10 B (−0.2%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.3 KiB	6.3 KiB	−1 B (−0.0%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	51.8 KiB	51.8 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	16	16	0 (0.0%)	21	✅
Claude	Total thread wire	12.6 KiB	12.6 KiB	−19 B (−0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.3 KiB	6.3 KiB	−4 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.3 KiB	6.3 KiB	−15 B (−0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	52.7 KiB	52.7 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	16	16	0 (0.0%)	21	✅

Baseline: d484735 · PR result: f3359b9 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 99.9 KiB
• Claude decoded thread snapshot: 100.6 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

Dismissing prior approval to re-evaluate f3359b9