Use case
A Claude Code dashboard for the whole team, live
Usage reports count tokens and cost. A team lead usually wants something else first: which threads are running right now, what they're on, and what's stuck, waiting on whom. AgentMingo shows that live, and counts a few things each week from the threads' own reports.
Free during the open beta. Works with Claude Code today; Codex and more are next. On macOS or Linux.
Right now: every reporting thread
The Sessions page, switched to everyone. Each thread shows its person, title, repository and branch, its state (Working, Needs your answer, Replied — your turn, Ready, Ended), its task, its pull request's status and its tokens. Overlapping files between threads are flagged.
Each week: what AgentMingo did, and how long things took
acme · your team's threads, last 7 days
58 threads worked
- Review results sent back
- 46
Up 7 on last week
- Review comments and CI failures you didn't have to copy-paste into a thread.
- Result to resolved
- 44mmedian, over 31 results
Up 8m on last week
- How long a review round takes, from the thread getting a result to it being resolved.
- Needed someone
- 19
Up 2 on last week
- Times a thread stopped on a prompt or a question, and you were told.
- Going again after a notice
- 10mmedian, over 19 waits
Up 3m on last week
- How long a thread waits on you once you've been told.
- Told before merge
- 12
Up 3 on last week
- Merge conflicts and main changes a thread heard about while it was still working.
- Steps handed
- 64
Up 17 on last week
- Workflow steps given to a thread one at a time, so you don't hand them over yourself.
- Review results sent back
- Review comments and CI failures you didn't have to copy-paste into a thread.
- Result to resolved
- How long a review round takes, from the thread getting a result to it being resolved.
- Needed someone
- Times a thread stopped on a prompt or a question, and you were told.
- Going again after a notice
- How long a thread waits on you once you've been told.
- Told before merge
- Merge conflicts and main changes a thread heard about while it was still working.
- Steps handed
- Workflow steps given to a thread one at a time, so you don't hand them over yourself.
Per task: what it took
A task's page adds up the threads tied to it: their tokens and cost at API list prices, the automatic review rounds on its pull requests, and the time from its first thread to its last merge. Cost is shown at API prices because that's what can be counted per thread. A subscription isn't billed per token, so this isn't your bill.
Your team's rules, in every thread
An admin writes the team's rules once, and every new thread gets them at its start, as guidance that never outranks the person at the keyboard. Built-in git habits (a short summary before a full diff, short logs, no lockfile diffs) go to new threads too, on by default. We don't claim they save anything until a fair test says so.
What it isn't
It isn't billing. For invoices and organization-wide usage, use your Claude plan's usage pages, the Anthropic Console, or Claude Code's OpenTelemetry export[1]. What AgentMingo sees is listed in the Privacy Policy: not your code or the conversation, except each thread's short title, what a waiting prompt asks (with answering on: its command, file path, question or plan), what the agent writes when it reports a workflow step, and the files it attaches.
Sources
Checked November 3, 2026, on each product's own pages. Products change: the links say what they say today.
- Claude Code: Claude Code exports usage, cost and activity through OpenTelemetry once telemetry is turned on. code.claude.com
Jira, Linear, GitHub, Shortcut and other product names on this site are trademarks of their owners. AgentMingo isn't affiliated with or endorsed by them.
Questions
Is the cost my real bill?
No. It's each thread's tokens priced at API list prices. On a subscription you aren't billed per token; the figure is there to compare threads and tasks.
Who can see what?
Everyone in the organization sees each thread's title, branch, state and counts. A waiting prompt's details, and answering it, are for the thread's own person only.
Does it replace OpenTelemetry dashboards?
No, they do different jobs. Claude Code's OpenTelemetry export sends usage, cost and activity metrics to a monitoring backend you choose. AgentMingo is a view of the work itself: which thread is on which task, its pull request, and what's waiting on whom.
How are the weekly numbers counted?
From your threads' reports and from what AgentMingo did, never estimated. A median is shown only over at least 5 cases. How we measure explains each one.
See what your team's coding agents are doing, live.
Sign up, run one command on your computer, and keep working. Free during the open beta.