Dashboard & Runs

Dashboard

The Dashboard is the health overview. The workspace dashboard shows, over the last 30 days, headline tiles (runs, success rate, spend, average duration), a trend, the top-failing agents, and recent improvements. Each agent also has its own dashboard with the same shape scoped to that agent — useful for deciding whether a model downgrade held up or an agent started failing.

Runs

The Runs page is the full, filterable history of every run in the workspace. Filter by agent, status, and trigger to find what you’re looking for, then open any run for its detail page — output, tokens, cost, and tools used. The “Action needed” alerts in the sidebar (failing agents, missing connections, missing LLM key) link straight to the relevant surface.