Open source, Apache 2.0
See what your AI agents are really doing.
Watch Claude Code and its subagents work, live in your browser or beside the chat. See what it costs, how close you are to your limits, and when it needs you. AgentMon Live also checks Claude’s work before it says “done”.
/plugin install agentmon-live --marketplace codenotary/agentmon-liveRun it in a Claude Code session (2.1.308 or later). There is nothing to set up.
The done gate
It checks before Claude says “done”
Claude says all tests pass? AgentMon Live checks, and sends it back if they don’t.
The check runs when Claude is about to stop after real work. Anything it finds goes back to Claude once per prompt: finish it, or say plainly why not.
- A failed, stale or skipped test, lint, typecheck or build
- A “tests pass” claim with no test run behind it
- A test run that ran no tests, or fewer than before
- Open todos, or a subagent that reported failing tests
- When the facts show nothing, a second model may check that every part of your request got an answer and that nothing you did not ask for got built
The web view
A live view that sits next to your terminal
Press o, or run /agentmon open, and your session opens in the browser. Put it on a second screen or beside your terminal: see who started whom, what each agent does right now, and where the money goes.
- One page: Overview, Tool calls, Verification, Subagents, Files and Session log
- Dials for context, quotas and cache hits, with cost, turns and time blocked on you
-
Runs at
http://127.0.0.1:47311/, on your machine only - Read-only: the one thing it can change is showing or hiding the pane
-
Needs
nodeon yourPATH; everything else works without it
Why you want it
See it happen
Every agent, tool call and handoff, live in your browser, in the agent flow.
Don’t run out
Cost, context and quota limits at a glance, and when /clear or /compact would save you money.
Know when to step in
It raises a waiting approval, a stuck subagent or a failure loop. Press i and it drafts the follow-up question for you.
Keep keys off GitHub
Clicked “don’t ask again” on a curl with your API token? It now sits in plain text in .claude/settings.local.json. AgentMon Live says so, in red when git would commit the file.
It learns your project
It remembers flaky tools and recurring errors across sessions, and spots drift in your CLAUDE.md.
Measure it
/agentmon bench compares your sessions with and without AgentMon Live.
The pane, beside the chat
It opens by itself when the terminal is wide enough, or with /agentmon.
Where your session data goes
The plugin reads the session, the done gate answers Claude, and a local server feeds the browser.
advisorMode to gate for none), a GitHub release check, and feedback only when you send it.One developer, one machine
Built for a single developer. Free, and yours to change.
AgentMon Live is for you and your own sessions: no account, no server to run, Apache 2.0 licensed.
Running more agents than one page can hold, or want to know how good their work is over time? AgentMon Start gives you a combined view of every agent on your machine, with a scorecard for cost, quality and security.
Install it in one line
/plugin install agentmon-live --marketplace codenotary/agentmon-live
Apache License 2.0.