Context Usage, Audit Log, and File Attachments
What’s New
Context Usage Breakdown See exactly what fills an agent’s context window on every run. A new usage panel attributes input tokens across instructions, conversation history, tools, Skills, and integrations, and an expandable context chip highlights the breakdown at a glance. Usage is aggregated across the runs in a session. See observability for details.
Audit Log A new audit log page records activity across your project, with filters by resource type and identifier so you can trace exactly what changed and when.
File Attachments for Agents Drag and drop multiple files straight into an agent run, or use the new attachment button. Uploaded files are tracked with their source and appear inline against the message that introduced them, in both single runs and multi-turn sessions.
Improvements
- Paginated agent run history plus a single-run detail view
- Visibility indicators on agent cards
- More reliable eval judging, with model parameter defaults applied consistently
- Smarter handling of uploaded files across an agent’s tools, so each tool receives only the files it needs
- Function names now appear in streaming events for clearer run traces
- A memory compaction indicator in the context usage panel
The context usage breakdown makes it easy to spot what is driving token cost on a run. If a single integration or Skill dominates your input, you can see it directly and trim what the agent carries.