Resend Lets Apps and Agents Check Their Own Quota in One Call
Resend's new GET /usage endpoint returns plan limits and current usage for email, contacts, domains and AI credits, with a CLI command and an MCP tool.
Resend now lets any API key read the account’s plan limits and current usage with one GET /usage request, the company said in an Oct. 1 changelog entry. Resend listed four uses: alerts before a quota runs out, pausing non-essential sends near a cap, internal dashboards and AI agents that track what they have left.
One request covers every quota
The response reports usage and limits for email, contacts, segments, broadcasts, domains, AI credits, automation runs and the account’s rate limit, according to Resend’s example. Email is split into daily and monthly windows, each with sent and received counts and a reset time. AI credits and automation runs carry their own reset or next-increase fields.
A limit of null means the plan has no cap on that resource, Resend said. Any API key can call the endpoint.
The CLI and MCP server get it too
Resend said the same data is available from the command line. resend usage prints a table. resend usage --json returns output for scripts. The company’s MCP server exposes it as a get-usage tool. An agent connected to Resend can read its remaining quota without a custom wrapper.
Resend published code samples for Node.js, PHP, Python, Ruby, Go, Rust, Java and .NET, plus curl.
Key takeaways
GET /usagereturns current usage and plan limits in one response.- Any API key can call it, and a
nulllimit means no cap. resend usageand the MCPget-usagetool return the same data.
The take
We think the useful part is that quota becomes something a program can read. Until now, many teams learned about a limit when a send failed. That is the worst moment to learn it. The message that fails is often the one someone was waiting for.
The agent case is the sharper one. An agent that can read its own budget can slow down or stop before it fails. An agent that only learns the limit by being refused cannot.
Teams on Resend should add a scheduled check that calls GET /usage and alerts at a set share of each limit, such as 80%. Teams running agents should have the agent read the same numbers before any bulk send.
