feat(web): setup ends with a Workflow step for automatic settling

[incognitojam]

Automatic settling surprises people who don't work the way it assumes: threads move into the Settled shelf when their PR merges or after three days idle, and nothing in setup says so. Setup is the one place every new user passes through, so it now explains settling and lets people keep or turn off those rules before they reach the sidebar.

Setup gains a final Workflow step after Projects (and Preferences, when that appears). It has a short explanation, a static picture of the sidebar with a settled thread, and two switches: settle when the pull request merges, and settle after N days without activity. The step is named for workflow defaults in general so more of them can join it later. Which ones join will be decided from usage data and user feedback.

• The switches start from the first selected computer's current settings, so re-running /welcome shows the saved choice.
• Finish writes only to computers whose values differ. Turning inactivity settling back on keeps that computer's own threshold. A failed save leaves the user on the step.
• Computers whose server lacks automatic settlement are left out; when none support it, the step doesn't appear and the import steps finish setup as before. While the step follows, the import actions read Import & continue and Skip projects.
• Finishing records a setup-thread-settlement usage event with only the two choices. This bumps the fork client usage vocabulary to version 3.
• User guides, the telemetry and feature-usage docs, the differences overview and the fork feature ledger are updated.

Before	After

Validation

In a dev server with isolated state, through /welcome in the Browser panel:

• Turned off the merge rule by clicking its label, chose Finish: setup closed onto a new thread and settings.json held sidebarAutoSettleOnMerge: false.
• Re-ran /welcome: the step opened with the merge switch off. Turned it back on and inactivity off, chose Finish: sidebarAutoSettleAfterDays: null, and Settings → General showed merge on and inactivity off.
• Turned inactivity back on and finished: the settings file returned to defaults.
• Focused tests pass for the patch logic (settlement.logic.test.ts), the usage event allowlist (ClientFeatureUsage.test.ts), and the existing onboarding and import tests; web typecheck and the ledger check pass.

Not verified: delivery of the usage event to PostHog (dev telemetry has no collector; the server test covers what is recorded), and the multi-computer path, which shares the per-computer patch logic covered by tests.

---

Written by an agent (Claude Code, claude-opus-5-5).

[github-actions]

Upstream divergence

Both sides are compared with upstream at the in-order tip 0ecb25a8d4. Excludes .repos/ and pnpm-lock.yaml.

Measure	Base	This PR	Change
Upstream files the fork changes	1,517	1,517	0
Lines changed in them	78,306	78,347	+41
Upstream files the fork removes	43	43	0
Fork-only files	558	561	+3
Fork-only lines	102,256	102,537	+281

Upstream files whose difference changes (2)

File	Change	Now differs by
apps/web/src/components/onboarding/WelcomeWizard.tsx	+34	986 lines
docs/user/welcome-wizard.md	+7	131 lines

[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	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	20.9 KiB	20.9 KiB	0 B (0.0%)	29.3 KiB	✅
Codex	Live turn messages	2	2	0 (0.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.2 KiB	21.2 KiB	0 B (0.0%)	29.3 KiB	✅
Claude	Live turn messages	2	2	0 (0.0%)	8	✅

Baseline: 7ce499e · PR result: e813284 · 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: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

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