How do I track usage per developer in Cursor (beta)?

Rius shows Cursor (beta) usage per developer by tracing each agent chat under the email that developer signed in with. Each developer installs the Rius plugin, signs in, and turns tracing on for a folder. Chats then appear per developer on the Users page, with their traces and failures. Cursor (beta) does not report token counts, so there are no tokens or cost per developer.

How it works

The Rius plugin turns each agent chat in Cursor (beta) into one trace. The trace holds the turns, the model for each answer, every tool call and the subagents the chat started.

• Who sent it. From plugin 0.6.2, every span carries user.id, set to the email the developer used at /rius-login. That email is what groups chats by developer.
• Opt in per developer and per folder. Installing sends nothing. Tracing stays off until a developer runs /rius-enable-here in a folder, and the agent never runs these commands on its own.
• What a chat records. With /rius-enable-here you get the span tree, timing, status, model names, tool names, error types and subagents. /rius-enable-content-here adds prompts, answers, and tool inputs and outputs, with recognised secrets removed. Redaction is best effort.
• When a chat counts. A chat’s trace closes when Cursor ends the session. If it never does, the developer’s next chat closes it once it has been idle for 12 hours. A chat counts in the Users and Agents figures once its trace is closed.

Cursor (beta) support needs plugin 0.6.0 or later. It is tested mostly against headless cursor-agent -p runs, so details may change.

Set it up

Each developer does this once on their own machine. The full install, sign-in and folder rules are in the Cursor (beta) guide.

1. Install the plugin. In Cursor’s plugin settings, choose Import from Repo and enter glassflow/rius-coding-agents. It needs Python 3.9 or later on PATH, and macOS or Linux.
2. Sign in to the shared workspace. Run /rius-login in the agent chat. In the browser, pick the team workspace and click Connect. Use one workspace for everyone so all developers appear in one Users list.
3. Turn tracing on for a folder. Run the enable command in each repo you want traced, then check the state.

/rius-login
/rius-enable-here
/rius-status

4. Pick a content level. Use /rius-enable-content-here only in repos where your team agrees that prompts and answers may be sent. Everyone in the workspace can read what is sent. /rius-disable-here stops a folder, and a disabled folder always wins over an enabled parent.
5. Check the Reason line. If /rius-status says no path rule matches, run /rius-enable-here in that folder. If it says there is no key, run /rius-login.

Do not store a shared key with use-key for this. A chat sent with a key stored that way carries no user and lands in the anonymous row.

What you get

Here is where per-developer usage shows in Rius for Cursor (beta) chats:

• Users page. Each developer appears by sign-in email, with their traces and failed traces for the selected time window. Open a developer to see their recent and failed chats. See Users for the view itself.
• Trace list. One row per chat. The Sent by column shows whose key sent each chat, and you can filter on it to one developer.
• Agents page. Chats from Cursor (beta) appear as an agent named cursor, with its traces and errors. The plugin sends no heartbeat, so nothing appears under Instances.
• Per chat detail. Open a chat to see the order of turns, answers, tool calls and subagents, and which tool call failed. A shell command that exits non-zero is marked as an error.

What Cursor (beta) does not report:

• Tokens and cost. Cursor’s hooks give plugins no usage data, so these traces carry no token counts and no cost, per developer or in total.
• Some sessions. Cursor’s cloud and background agents and Tab completions are not traced.
• Exact timing. Span times are when Cursor reported each step, not when the model call started.

You can also ask about your own chats in plain language through the Rius MCP server the plugin registers as rius. It uses a second sign-in, separate from /rius-login.

When something else fits

If you need tokens and cost per developer, Claude Code and Codex (beta) report them. See Track Claude Code usage and cost across a team. To compare all the supported agents side by side, see Coding agents.

Related