Agent Activity

Workbench | Preview Enhanced

WarningPreview feature

This feature is in preview. Preview features are unsupported and may face breaking changes in a future release. Any issues found in the feature will be addressed during the regular release schedule; they will not result in immediate patches or hotfixes.

We encourage customers to try these features and we welcome any feedback via Posit Support, but we recommend that the feature not be used in production until it is in general availability (i.e., officially released as a full feature). To provide feedback, please email your Posit Customer Success representative or and specify that you are trialing this feature.

Agent Activity gives you a view of your AI coding-agent usage on the Workbench homepage, including recent sessions, usage totals, estimated costs, models, tools, and workflow patterns.

The page shows your own activity. It does not rank users or provide an organization-wide administrator dashboard.

Open Agent Activity

When your administrator has enabled the feature, select Agent Activity in the header of the Workbench homepage.

Current Workbench homepage with a red box around the Agent Activity tab in the header navigation, next to Projects and Jobs.

The Workbench homepage with the Agent Activity tab highlighted.

If the tab is absent, ask your Workbench administrator whether Agent Activity is available in your deployment. Administrators can find setup instructions in Agent Activity in the Admin Guide.

The page shows Starting while the service starts and indexes your transcript history. The first start can take several minutes. See If Agent Activity does not load if the page stays in this state.

Activity

The Activity tab shows summary cards and recent coding-agent sessions. The cards summarize active time, session count, distinct projects and models, and total estimated cost. Recent sessions show the project, start time, duration, and estimated cost when available.

Agent Activity page with usage summary cards, Activity, How you work, and Quality tabs, and a list of recent coding-agent sessions.

Agent Activity with usage summary cards and recent sessions.

How you work

The How you work tab summarizes patterns such as Plan Mode use, subagents, skills, pacing, model spend, tool calls, and session shape.

How you work tab with workflow metric cards, model and tool usage charts, and session-shape distributions.

The How you work tab with workflow and session-shape analytics.

Quality

The Quality tab presents signals derived from your session history:

  • Cache efficiency: how much of your context the cache served and what that saved
  • Rework and friction: areas that negatively impact efficiency
    • Where work stalled on tool retries
    • Context compactions
    • Repeated edits
  • Prompting habits: suggestions for sharpening your prompts

Quality tab with a Cache efficiency card showing cache hit ratio and savings, a Rework and friction card with bars for tool retries, compactions, and edit churn, and a Prompting habits card listing three suggestions.

The Quality tab with cache efficiency, rework, and prompting-habit signals.

Your session history

Agent Activity reads supported coding-agent transcripts from your home directory, including Claude Code history under ~/.claude/projects/, Codex history under ~/.codex/sessions/, and Posit Assistant history under ~/.posit/assistant/workspaces/. Enabling Agent Activity does not install or enable a coding agent.

If the page has no activity, allow the initial indexing operation to finish and confirm that you have supported coding-agent history in your Workbench home directory.

Agent Activity page with zeroed summary cards and a Recent Sessions list that reads: No agent sessions in the last 30 days yet.

Agent Activity with no sessions in the last 30 days.

If your history still does not appear after a few minutes, contact your Workbench administrator.

If Agent Activity does not load

Agent Activity might require a long, one-time setup period. Usually this occurs if Workbench has been restarted by the admin. While it starts, the page shows Starting with a progress indicator in place of your activity:

Agent Activity page with an AgentsView card that reads Starting and explains that the first start can take a couple of minutes.

Agent Activity while the service is starting.

The card can show a more specific message from Workbench in place of the default text. Leave the page open. It updates automatically when the service is ready.

If the service cannot start, the page shows Not running:

Agent Activity page with an AgentsView card that reads Not running and suggests contacting your administrator if it stays unavailable.

Agent Activity when the service is not running.

The card can include a Last reported status line with the reason Workbench recorded. Reload the page to try again. If the page stays on Starting for more than 15 minutes, or keeps showing Not running, contact your Workbench administrator and include any status message shown on the card.

Back to top