01 Agent usage monitor
Know what's left.
Keep moving.
Real-time Codex and Cursor limits in a native, local-first monitor. No dashboard detours. No second CLI. No guessing.
Monitoring Codex · Codex + Cursor available at launch
Usage
The answer is one glance away.
03 The hidden cost
The expensive part isn't checking. It's coming back.
Agents move quickly. Usage information should not ask you to leave the work, open another surface, or remember how each provider reports limits.
Re-open the provider
Find the right account, tab, and tiny usage indicator again.
Guess the remaining runway
A limit is only useful when it is visible before the interruption.
Lose the thread
Context switching costs more than the check itself.
04 Provider ledger
Different limits. One readable system.
useagent normalizes the signals that matter and keeps them close to the work.
Codex
OpenAI's terminal code agent
Cursor
AI-first editor, in your flow
05 The system
Quiet by default. Useful on demand.
Read
Listen for provider state where it already lives.
Normalize
Turn different windows and limits into one language.
Surface
Put the answer within reach, without another destination.
Headroom remaining
Across the active Codex window, there is still room to keep the thought moving.
checked 08:42:17 UTC · scope: edge07 Early access
Keep the thread. We'll keep the count.
useagent is being shaped with people who live in their coding agents. Join the small first wave.