See each agent’s reported usage, then hand work to the other agent in a new thread.
| Agent | Where to check and what resets |
|---|---|
| Codex | Usage dashboard or /status in the Codex CLI.Plus and Standard Business: five-hour estimates. Pro: no five-hour limit currently. Weekly limits may apply. Plan details. |
| Claude Code | /usage in Claude Code. Limit messages show reset times.Pro and Max: five-hour and weekly limits, shared with Claude. Plan details. |
Use Codex and Claude Code in bb on your own plans. If one reaches its limit, start a thread with the other when it has allowance left. Using the other agent doesn’t reset either provider’s limits.
Plans and windows can change. API-key accounts follow their own billing and limits. bb shows the windows your plan reports.
Ask me about each of these, one at a time, with a suggested default: - Machine (the computer running your task) - Agents (Codex and Claude Code accounts you already use) - Task (what you want to continue if one agent is limited) Skip any I've answered. Then sum up the plan in three bullets and wait for my go. My answers override the steps below. If a wait times out, wait again, up to three times. Help me check my agents’ reported usage and prepare a handoff if one has hit a limit. Guide: https://getbb.app/guides/codex-claude-code-usage-limits Do not buy credits, change accounts, change retry settings, or start another agent turn without asking. Do not read credential files or print account emails or tokens. If a check fails, stop and report the command or action, the error, and the next move. 1. Identify the bb server, project, environment and machine for this task. Confirm that they match my answers. Do not change production settings. Check: state the machine and project before looking up usage. 2. Inspect the installed Provider usage plugin’s read-only RPC contract with bb plugin rpc list and bb plugin rpc inspect. Follow its published input schema to read usage for the selected machine. Do not use bb settings usage as an aggregate view, and do not inspect or configure Account Pooler. Check: report only the providers and usage windows actually returned, with percent used and reset time when present. Treat unavailable or unreported usage as unknown, never as zero or unlimited. If no supported read-only path is available, point me to Settings → Installed plugins → Provider usage and stop. 3. If the other agent reports available allowance, prepare a short handoff containing the task, project or checkout, completed work, relevant files, next step, and anything requiring approval. Do not claim the other agent inherits the conversation. Do not start it yet or let two agents edit the same checkout at once. Check: the handoff names the current state and a bounded next step without secrets. 4. Explain the choice: wait for the current provider’s reset, or start a new thread with my other agent and paste the handoff. Switching agents does not increase or reset either provider’s allowance. Check: ask before starting any new work, and leave accounts and settings unchanged. Reply with the machine, each reported window and reset time, any missing information, and the handoff ready to copy.
See the plan, percent used and reset time for each reported window.
Provider usage is built into bb and on by default for new installations; existing choices are kept. You can also check it from the sidebar’s Provider usage card. The screenshot shows example values in a local demo.
Use a new thread for the other agent, on its own account and allowance.
Continue this task from the current project files. First summarize what is done and what you would do next. Wait for my go before editing.
Some plans do not report usage windows. That is different from a failed refresh or an expired sign-in. Check the message in Provider usage and the provider’s own usage page. bb doesn’t show unavailable usage as zero.
Open Settings → Installed plugins → Provider usage and pick the computer running the task. The selected machine determines which local accounts you see.
Reconnect the selected machine and refresh Provider usage. After a failed refresh, bb keeps the last measurements with a notice; those figures are not a fresh allowance check.
bb picks the work back up. When an agent stops on a usage limit that reports when it resets, bb sends the message again a little after the reset, up to four times per turn, so you don’t have to come back and press send. Credit and spend limits aren’t retried.
By default bb waits up to 6 hours for a reset. For a weekly limit, raise Maximum automatic wait in Settings → Installed plugins → Provider retry. The machine running the agent needs to stay available.
Yes. bb runs the agents you already use, signed in the way you already pay for them: a Claude Pro or Max plan, a ChatGPT plan for Codex, a Cursor plan, API keys, or any other provider’s plan. Your existing setup comes along too (CLAUDE.md, skills, MCP servers, and agent settings).
No. Each provider keeps its own limits. Using both means starting work with another agent on an account you already have, if that account has allowance available. bb doesn’t increase or reset either limit.
No. As of October 21, 2026, OpenAI lists five-hour estimates for Plus and Standard Business, while Pro currently has no five-hour limit. Weekly limits may apply. Check your Codex usage dashboard and OpenAI’s current plan details.
Yes. Anthropic’s Pro and Max guidance says usage is shared across Claude and Claude Code, with five-hour and weekly limits. Check /usage in Claude Code for your plan.
This guide uses a new thread and an explicit handoff. Give the next agent the task, current files and next step; do not assume it can see the previous conversation.
Provider documentation was checked October 21, 2026: OpenAI’s Codex limits, Anthropic’s plan limits and Claude Code commands. bb’s display and retry behavior are documented in Provider usage and Provider retry.
Check the usage your plans report, then choose which agent takes the next task. The bb app is free and open source.