Summary

Type: SPIKE — re-audit prior gut areas for lingering UI remnants.

The recent UI remnants audit focused on thinking / skills / plugins / providers / onboarding / Pi-orchestrator. Adjacent gut areas from earlier PRs were NOT re-audited and may have left orphan state, types, or i18n in ui/src/.

Scope — gut areas to re-audit

• gut: remove session cost analytics (MBP audit) #2307 / gut(infra): remove session cost analytics dashboard (#2307) #2332 — session cost analytics dashboard (gutted). Any UI remnants (types, state, i18n, charts)?
• gut(session): remove vestigial model fallback system #2106 / gut(session): remove vestigial model fallback system #2130 — model fallback system (gutted). Any UI remnants beyond SECTION_META.models already flagged?
• gut(agents): replace default-agent display ordering with list-order semantics #2286 — default-agent display ordering (gutted). Any UI remnants?
• gut(session): remove vestigial contextTokens config and display denominator #2277 — contextTokens config and display denominator (gutted). Re-grep to confirm grep-clean.
• Dead LLM provider code + zero-caller stubs cleanup #2363 — LLM provider code + zero-caller stubs (gutted). Any UI coupling?

Approach

For each gut area:

1. Identify distinctive terms from the gut PR
2. grep -rn '{term}' ui/src/ — classify each hit as LIVE / ORPHAN / DEAD WIRING
3. Record findings in audit doc

Deliverable

Per-area report (either confirmed clean or a list of remnants with classification). For remnants, follow-up cleanup issue(s).

AC

• Per gut-area entry in the audit doc: either "grep-clean confirmed" or a classified list of remnants
• Follow-up cleanup issues opened for any remnants found

Typed exception

SPIKE — AC is the audit outcome.

Context

Post-#2336 UI remnants audit was focused on user-provided subsystems (thinking, skills, plugins, etc.). These adjacent gut areas were not in scope but likely have parallel patterns.