Favicon of Claude Code Agent Monitor

Claude Code Agent Monitor

A local AI agent dashboard monitors Claude Code, Cursor and Codex, with offline session history, cost estimates and browser-based agent controls.

Screenshot of Claude Code Agent Monitor website

Claude Code Agent Monitor gives developers one place to follow coding agents and inspect their work across Claude Code, Cursor and Codex. It keeps conversations and activity available for review after a session ends.

  • Live session monitoring: Follow tool calls, command output and agent status, with a Kanban board showing work that's active, waiting, completed or in error.
  • Searchable history: Replay conversations and inspect subagent activity, including Cursor history captured before the editor removes it.
  • Usage and workflow analysis: Compare model and subagent cost estimates, track tokens through context compression, and view delegation, concurrency and tool-flow diagrams.
  • Agent controls and configuration: Start or resume Claude Code and Codex sessions in the browser, send follow-ups, inspect settings and edit supported configuration files with backups.
  • Alerts and integrations: Rules can flag stuck agents, error bursts and token thresholds; optional webhooks, a local MCP server, Prometheus and Grafana connect monitoring to other tools.

The dashboard works fully offline, requires no account and sends no telemetry. Monitoring data stays in local SQLite storage by default, and the dashboard doesn't send transcripts to a hosted service. These facts apply to the monitor; running the connected agents follows each agent's own service requirements.

It's open source under MIT and runs as a local browser dashboard, a macOS or Windows desktop app, or a self-hosted Docker or Podman deployment. A VS Code extension provides an editor sidebar. Optional remote sources bring Claude Code and Codex histories in over SSH or authenticated push, while external webhook destinations receive alerts you choose to send.

Similar to Claude Code Agent Monitor