How do I see Claude Code usage across my whole engineering team?

Rius shows Claude Code usage and cost per developer when everyone signs in to one shared workspace through the rius plugin. Each developer installs the plugin, signs in with /rius:login, and turns tracing on for a repo with /rius:enable-here. Every session then appears as a trace, and the Users page lists each developer with their traces, failed traces and cost. Costs are list-price estimates in US dollars.

How it works

The plugin turns each Claude Code session into one trace. The trace holds the session as its root, each turn (one prompt and everything it caused), each model call with its tokens, cache reads and cost, each tool call, and each subagent with its own model calls, tokens and cost. Subagents can account for more than half of all tokens in long sessions, so seeing them separately matters for cost.

When a developer signs in, the plugin stores an API key for that machine and the workspace they chose. Every span then carries the developer’s sign-in email as user.id. That is how Rius groups sessions per person:

• The Users page lists each developer with Traces (one per session), Failed traces and Cost for the selected window, 7 days by default. Sort by Cost to see who drives spend.
• The trace list shows one row per session. The Sent by column names the owner of the key that sent it, and clicking it filters to one developer.
• The Agents page shows Claude Code as an agent named claude-code, with its traces, errors and cost, and each open session as a live instance.

Users and Agents count a session once it has ended. The trace list shows it while it runs. A session that never ends cleanly, because Claude Code crashed or was killed, is never counted on those two pages.

Set it up

1. Pick one workspace for the team. A workspace admin invites teammates under Settings → Organization. Teammates accept the invitation first, then pick the team workspace when they sign in.

2. Each developer installs the plugin. Run this in a terminal. It needs Python 3.9 or later on the PATH:

   claude plugin marketplace add glassflow/rius-coding-agents && claude plugin install rius@rius-coding-agents

3. Each developer signs in. In Claude Code, run:

   /rius:login

   In the browser at console.rius-glassflow.com, sign in or sign up, check that the page shows the same code Claude Code printed, pick the team workspace, and click Connect. The stored key only sends traces, expires after 90 days, and is listed under Settings → API keys as Claude Code: followed by the email and machine. An admin can revoke it there. Run /rius:login again when it expires.

4. Each developer turns tracing on per repo. Tracing is off until a developer enables a folder, and nobody can switch it on for the team from a repo’s settings:

   /rius:enable-here

   Then run /rius:status. The first line reads Rius tracing: on.

5. Start Claude Code from inside the enabled repo, not from a parent folder. A session started elsewhere can stop being traced partway through.

6. Agree which repos send content. /rius:enable-here sends structure only: the span tree, models, tokens, cost, timing, tool names and error types, plus the developer’s email, the folder path and the git branch. /rius:enable-content-here also sends prompts, replies, file contents and command output, with recognized secrets removed on a best-effort basis. Everyone in the workspace can read what is sent. To stop a folder, run /rius:disable-here. A disabled folder always wins, even inside an enabled parent.

7. Read usage. Open the Users page and sort by Cost. Open a developer to see their recent and failed sessions. Each developer can also run /mcp, pick rius, and sign in once to ask Claude questions such as which sessions cost the most this week.

The full plugin guide covers sign-out, environment variables and troubleshooting: Claude Code plugin. The team rollout is written up in Track Claude Code usage and cost across a team.

What you get

• One workspace where cost and usage for every developer add up.
• A per-developer view: traces, failed traces and cost on the Users page, and a Sent by filter on the trace list.
• A waterfall for each session, with tokens and cost on every model call and failed tool calls marked as errors, so you can go from “this session cost $12” to the exact step.
• Subagent tokens and cost shown on their own spans.
• Per-folder control over content. Developers choose structure only or full content, and RIUS_CAPTURE_CONTENT set to false in a Claude Code settings file turns content off everywhere on that machine.

Rius prices tokens at API rates. On a Claude Pro or Max subscription, the cost measures usage and is not what you pay. For more ways to trace coding agents, see Coding agents.

When something else fits

• A session sent with a key you stored yourself, or from an older plugin before 0.6.2, carries no developer and lands in the anonymous row on Users, so have everyone sign in with /rius:login on a current plugin.
• If your team also uses Codex (beta) or Cursor (beta), the same plugin covers them, but Cursor sessions carry no tokens or cost because Cursor’s hooks do not report usage.
• OpenHands needs no plugin. Three OTLP environment variables send its conversations to the same workspace, and you pass user_id to see each developer.

Related