fix: compact boundary marker showing "undefined · ~NaNK tokens" on codex sessions

[k11kirky]

Problem

A user on a codex model (gpt-5.6-sol) saw the compaction marker render as "Conversation compacted · undefined · ~NaNK tokens". The codex app-server adapter emits _posthog/compact_boundary with only sessionId, but the UI assumed trigger and preTokens (which only the Claude adapter sends) were always present, so it interpolated undefined and NaN straight into the marker.

Why

Codex sessions compact regularly, so every codex user hitting auto-compaction sees this broken marker in the transcript.

Changes

• CompactBoundaryView (and the item types feeding it) now treat trigger/preTokens as optional and drop the missing segments instead of rendering undefined/NaN.
• The codex adapter now reports trigger: "auto" (codex only self-compacts) and preTokens from its usage tracker's last context-occupancy reading, so the marker shows real data going forward.

How did you test this?

• vitest run codex-app-server-agent in packages/agent (75 passed), including an updated boundary test asserting trigger/preTokens.
• vitest run buildConversationItems in packages/ui (32 passed).
• tsc --noEmit on @posthog/agent and @posthog/ui; biome check on changed files (only pre-existing warnings).

Automatic notifications

• Publish to changelog?
• Alert Sales and Marketing teams?

---

Created with PostHog Code

[trunk-io]

Merging to main in this repository is managed by Trunk.

• To merge this pull request, check the box to the left or comment /trunk merge below.

After your PR is submitted to the merge queue, this comment will be automatically updated with its status. If the PR fails, failure details will also be posted here

[github-actions]

React Doctor found no issues in the changed files. 🎉

_{Reviewed by React Doctor for commit cb2f39e.}