Read Usage and Plan Limits

Part of Monitor.

On Your Machine

cedr monitor                  # the last 7 days by model
cedr monitor --since 30d --by project
cedr monitor --by day         # also: harness, hour, session, effort
cedr monitor sessions         # the costliest sessions
cedr monitor limits           # each plan window: use, reset, projection, outlook
cedr monitor --format json    # for scripts and agents

The first run reads your history and can take a minute; later runs read only what is new. The report’s totals include cost, tokens by type, responses, sessions, cache hit rate, net cache saving, tool calls and errors, API errors and estimated active time.

How spend is counted. Spend is what the same tokens would cost on each vendor’s API at list prices, not what your plan charges. Comparing it with what you pay shows how much your plan is worth to you. Models Monitor has no price for count as $0 and are marked unpriced.

Plan limits. For each window (a 5-hour or weekly limit), Monitor shows the use, the reset time, the use projected at reset from the pace so far, when it would run out, and an outlook: on track, tight (projected 70% or more), limit risk (projected 100% or more) or at limit. For Claude Code, its status line must run cedr route statusline (Set Up Route explains it); the setup prompt wires it for you.