Track Claude Code usage and cost across a team

Install the rius-coding-agents plugin in Claude Code, sign in to a shared Rius workspace, and turn tracing on for the repos you want to watch. Every session then streams to Rius as an OpenTelemetry trace with its model calls, tokens, cost, tool calls, and subagents, per developer and per session. Prompts and replies are sent only from folders where a developer turns content on. The same plugin covers Codex and Cursor in beta, and OpenHands sends to the same workspace over OpenTelemetry.

Set it up for your team

Pick one workspace for the team

Use one Rius workspace for everyone’s coding sessions, so cost and usage add up in one place. A workspace admin invites your teammates under Settings → Organization. Teammates accept the invitation first, then pick the team workspace when they sign in below.

Each developer installs the plugin

In a terminal:


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

The plugin needs Python 3.9 or later; see Requirements. To install from inside Claude Code instead, see Install.

Each developer signs in


/rius:login

In the browser, sign in, pick the team workspace, and click Connect. The plugin stores a key for that machine and workspace.

Each developer turns tracing on for a repo

In each repo you want traced:


/rius:enable-here

Tracing is opt-in per developer and per folder. You can’t turn it on for the team from a repo’s settings, and Claude can’t turn it on itself: each developer runs the command. /rius:status confirms the state: the first line reads Rius tracing: on.

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

What each session records

One Claude Code session is one trace. Its spans fill in while the session runs:

The session, as the root of the trace.
Each turn: one prompt and everything it caused.
Each model call, with the model, input and output tokens, cache reads and writes, and cost.
Each tool call: Bash, Read, Edit, MCP tools, and so on, with errors marked on the span.
Each subagent, nested under the call that started it, with its own model calls, tokens, and cost.

A folder enabled with /rius:enable-here sends structure only: the span tree, models, tokens, cost, timing, tool names, and error types, plus the developer’s email (as user.id), the folder path (which can include the local user name), and the git branch. It sends no prompts, replies, file contents, or command output. To also send prompts, replies, file contents, and command output, enable the folder with /rius:enable-content-here instead. The plugin removes the secrets it recognizes first, on a best-effort basis. Everyone in the workspace can read what is sent, so agree as a team which repos send content. What gets captured has the full list.

Rius prices tokens at API rates. On a Claude Pro or Max subscription, the cost in Rius measures usage; it isn’t what you pay.

See usage and cost per developer

With plugin 0.6.2 or later, each session is recorded under the email the developer signed in with, as user.id. In the console:

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

The Agents page with the claude-code agent: one live session and two ended ones, and the window's traces and cost

The Users and Agents figures 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 there.

Sessions from older plugin versions, and sessions sent with an existing API key instead of /rius:login, carry no developer and land in the anonymous row on Users.

See what one session is doing

Open the session in the trace list. The waterfall shows every turn, model call, tool call, and subagent in order, with tokens and cost on each model call. A failed tool call is marked as an error, so you can go from “this session cost $12” or “this session went wrong” to the exact step.

A Claude Code session in the console: the span tree with model calls and a failed Bash tool call, and one model call's output

You can also ask from the terminal. The plugin brings the Rius MCP server. Run /mcp, pick rius, and sign in with your Rius account once. Then ask Claude questions like:

“Which Claude Code sessions in the last 7 days cost the most, and what did the tokens go on?”
“Show me my last failed session and which tool call went wrong.”
“How much did subagents cost this week compared with the main sessions?”

See Query your sessions.

Codex (beta)

The same plugin traces the Codex CLI: one trace per session, with model calls and their tokens, tool calls, and subagents, under an agent named codex. It installs with:


codex plugin marketplace add glassflow/rius-coding-agents && codex plugin add rius@rius-coding-agents

Codex keeps its own sign-in and folder rules. See Codex (beta).

Cursor (beta)

The same plugin traces Cursor agent chats, in the editor and in cursor-agent: one trace per chat, with tool calls and subagents. Cursor doesn’t give plugins token counts, so Cursor traces show no tokens and no cost. See Cursor (beta).

OpenHands

OpenHands needs no plugin. Its SDK already traces every conversation with OpenTelemetry, so three environment variables send each conversation to Rius as one trace, with model calls, tokens, cost, and tool calls. To see each developer on the Users page, pass user_id when you create the conversation; see Attribute conversations to a user.

Claude Code plugin: the full guide, including what gets captured, signing out, and troubleshooting.
Users: the per-user view of traces, failures, and cost.
Settings: workspaces, members, and API keys.