Follow up to: #3573715 ## Problem/motivation **Per-item visibility (recommended for rc1 if feasible):** * Show the **token size** of each Context Item. This helps site builders understand the cost and context-window impact of each item, especially as items grow or include large File / Connected sources. * Show **usage counts** so a site builder can quickly see which items are being called frequently and which are rarely used. **Usage over time (nice to have):** * Add a **table of usage by hour** (or by day) for each Context Item. The purpose is to let site builders see whether a Context Item is being called more or less than expected, which helps catch misconfigured agents, runaway loops, or items that aren't being picked up at all. * Designed as a table for rc1 simplicity, with a clear path to becoming a chart in a future release once the data shape is validated. **Followup review:** Usage and logging screens not yet reviewed in detail. Recommend a follow-up review session focused on: * How usage metrics are presented (per Context Item, per agent, per use case) * Log filtering and search affordances * Connected Context sync history and error reporting **Additional discussion:** * Long-term: surface live token usage in the chat / agent run interface ## Solution TBD Given we will be moving to observability integration in the coming months, we should sort these out at the same time. ## Tasks - Discuss with observability in mind - Make decisions - Move forward with agreed upon improvements - Update code - Review and test ## AI usage disclosures - [ ] AI assisted issue - [ ] AI assisted code - [ ] AI generated code