[tusharbhardwaj-bk]

Follow-up to #139/#140.

What changes

Clicking an agent view's title expands it to fill the message area. The composer
stays exactly where it is, so you can keep talking to the agent while the view is
open. Escape or Close returns it to the inline box.

The inline box keeps its own Hide/Show collapse toggle, so the two
gestures don't fight: the title expands, the right-hand control collapses.

Why it is built this way

The overlay mounts as the last child of the messages wrapper in ChatView
rather than inside the timeline row that opened it. Timeline rows are
virtualized — they clip their children and get recycled — so an absolutely
positioned overlay inside one would be cut off or vanish on scroll. The messages
wrapper is already relative and already excludes the input bar, so
absolute inset-0 there covers precisely the transcript and nothing else, which
is what was asked for.

Which view is expanded lives in a small fork-owned zustand store rather than in
the row, which is what lets the card and the overlay be different mount points.
It is deliberately single-valued: expanding a second view replaces the first,
because one overlay mount cannot stack.

The iframe itself moved into a shared AgentUiRenderFrame used by both the card
and the overlay, so sandboxing, loading and failure states cannot drift between
them — the sandbox rules especially.

Escape is handled on the capture phase: the composer and timeline both handle
Escape, and the frontmost surface should answer first.

Fork surface

One new fork-owned file (agentUiExpandedStore.ts), the rest inside the existing
fork/agentUiSurface.tsx. ChatView takes one marked import and one marked
line.

Verification

• agentUiExpandedStore.test.ts — 4 tests: initial state, expand/collapse
  round-trip, replace-not-stack, idempotent collapse.
• Existing 9 surface tests still pass, including the sandbox-origin boundary.
• Web fork suites: 132 tests green. Typecheck clean under tsgo, lint clean,
  format clean, fork markers pass.

Written by Claude Opus 5 in T3 Code.

🤖 Generated with Claude Code

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

ℹ️ No successful main baseline artifact is available yet. This run establishes the initial measurement.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	—	13.6 KiB	—	15.1 KiB	✅
Codex	Thread snapshot wire	—	7.2 KiB	—	7.3 KiB	✅
Codex	Live turn WebSocket wire	—	6.4 KiB	—	7.8 KiB	✅
Codex	Live turn WebSocket decoded	—	55.7 KiB	—	66.4 KiB	✅
Codex	Live turn messages	—	11	—	21	✅
Claude	Total thread wire	—	13.6 KiB	—	15.1 KiB	✅
Claude	Thread snapshot wire	—	7.2 KiB	—	7.3 KiB	✅
Claude	Live turn WebSocket wire	—	6.4 KiB	—	7.8 KiB	✅
Claude	Live turn WebSocket decoded	—	56.5 KiB	—	66.4 KiB	✅
Claude	Live turn messages	—	11	—	21	✅

Baseline: unavailable · PR result: 8204ea5 · 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: 110.4 KiB
• Claude decoded thread snapshot: 111.1 KiB

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