1. Read remaining usage and reset time together
A low percentage may be manageable when reset is close, while a healthy-looking percentage can still be risky if the weekly reset is days away and the next task is large. Separate the active session allowance from the longer weekly allowance.
2. Make tasks smaller in the risk zone
- Finish work at a file or feature boundary before starting a broad refactor.
- Save and verify current changes before opening a long new context.
- Keep research, documentation, and testing tasks available for a tool switch.
- Near reset, favor review and cleanup over opening a large unfinished branch.
This does not increase quota. It reduces the amount of unfinished work left behind when a limit interrupts the session.
3. Lower the cost of checking
Opening several dashboards can become its own interruption. RunDog uses runner speed for routine awareness and keeps exact Claude and Codex values in one popover. Usage refreshes every 60 seconds, pauses while the Mac sleeps, and can be refreshed manually.
- Normal work
- Glance at the runner
- Runner speeds up
- Open remaining usage and reset time
- Limit reached
- Move to a prepared task until reset
What the current version does not claim
RunDog currently shows provider-reported remaining usage and reset times. It does not predict an exhaustion time from your recent burn rate. The interface avoids presenting an estimate as a measured fact.
4. Plan the day around reset times
The session allowance resets on a short cycle, while the weekly allowance resets on a fixed day and time. A single look at the popover countdown is enough to split work into "finish before this session ends" and "start after the reset". Spend the last stretch before a weekly reset on review and cleanup, then place the largest task right after the reset — the number of sessions cut short by a limit drops noticeably.
5. Keep a ready list for when the limit hits
Deciding what to do at the moment the runner falls asleep breaks the flow. Prepare a short list of work that moves forward without the assistant:
- Read through the changes you just made and tidy the commit message.
- Run the test suite and note any failing cases.
- Write requirements and design notes for the next task.
- Groom the issue tracker or walk through the release checklist.
When the reset passes and the runner starts running again, leave the rest of the list for next time and return to the original task.
6. Running two providers shortens the sleeping time
If you use both Claude and Codex, connect both accounts in RunDog and keep two runners side by side. Give each runner a different color in the settings window so the service that sped up is recognizable by color alone. When one runner falls asleep, move work to the other provider and come back after its reset; alternating between the two allowances greatly reduces the time you are fully stopped.
- Two runners
- Claude and Codex remaining usage shown separately
- Color coding
- Per-service presets, rainbow, or accent color
- Alternate
- Switch providers when one runner sleeps
Options for a quieter menu bar
When you need focus, the settings window can stop the runner for a while; when you want variety instead, a shuffle option switches to a different runner every 10 minutes. The System tab shows CPU, memory, storage, battery, and network together, so you can tell in one place whether a slow response is caused by a limit or by load on the Mac itself. System values are measured only while the panel is open and never leave the device.
Frequently asked questions
- Does it refresh during sleep?
- No. Polling pauses while the Mac sleeps and resumes on the 60-second cycle after wake.
- Does it predict when usage runs out?
- Not yet. It shows only the provider-reported remaining usage and reset time.
- Does it cost anything?
- The app is free and needs no separate account or subscription.