FernandoTechSpace

Jarvis AI Monitor Omarchy — AI skill for Claude Code

AI community

AI observability plugin for the Omarchy shell: see which AI coding agents are on the machine, what they are connected to, what they are allowed to do, and what they did last.

How to install Jarvis AI Monitor Omarchy

This entry records only its repository, not the path inside it, so there is no exact command to give. Open FernandoTechSpace/jarvis-ai-monitor-omarchy and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Jarvis AI Monitor Omarchy does

AI observability plugin for the Omarchy shell: see which AI coding agents are on the machine, what they are connected to, what they are allowed to do, and what they did last.

Alternatives in AI

Invisible Dots — Open-source, self-hosted alternative to OpenAI Dots, Meta Muse, Grok Bot, Manus Cue and Claude Cowork 31.8k ★
Claw Empire — Command Your AI Agent Empire from the CEO Desk — A local-first AI agent office simulator that orchestrates CLI 1.1k ★
Kibitz — Real-time decoded feed of AI agent actions — monitor multiple Claude Code & Codex sessions, see exactly what e 521 ★

README

Jarvis AI Monitor

AI observability plugin for the Omarchy shell: a local monitor for the AI coding agents on the machine. It puts one icon in the top bar and opens a panel that answers four questions about them: which ones are installed, what they are connected to, what they are allowed to do, and what they did last.

No sudo or pkexec is required. The plugin itself opens no network connection.

It is observability for AI agents kept on the machine itself: sessions, network connections, MCP servers, permissions and activity are read from local files and shown in the bar, with no telemetry, account or external service.

![Jarvis AI Monitor panel](preview.png)

Bar icon

What it shows

The panel lists every agent found, and for the selected one:

Section	Content
Last action	the most recent tool call or reply, when it happened and in which folder
Sessions	each running session: pid, name, busy or idle, uptime, version, working folder, permission mode, its own last action
Network connections	TCP and UDP sockets open by the session and by every process it spawned, grouped by remote endpoint
MCP servers and connectors	configured servers with type, target and the scope that declares them
Permissions	default mode, allow, ask and deny rules per scope (managed, user, project), trusted folders

Agents detected:

Agent	Binary	Detail level
Claude Code	`claude`	sessions, permission mode, rules, MCP, connectors, last tool call
Codex	`codex`	approval and sandbox policy, trusted projects, MCP, last thread
opencode	`opencode`	permission config, MCP, last activity time
Gemini CLI	`gemini`	allowed and excluded tools, MCP, last activity time
Cursor Agent	`cursor-agent`	allow and deny rules, MCP, last activity time
Crush	`crush`	allowed tools, MCP, last activity time
GitHub Copilot CLI	`copilot`	trusted