AgentsMonitoring

Agents Monitoring

  

Keep your AI coding agents alive and watch their status — **Claude Code**, **Codex**, and background daemons like **OpenClaw** or **Hermes**.

Agents Monitoring is a tiny, dependency-free supervisor. It **auto-detects** what's running in your tmux sessions (you don't declare anything — it looks), shows a **live status page**, and **restarts** anything that dies — surviving logout and reboot.

tmux sessions  ⇄  Agents Monitoring  ⇄  status page + keepalive

---

Why it's built this way

• Auto-detect, don't configure — it enumerates tmux sessions, walks each process tree, and classifies what's running (Claude Code / Codex / …). Setup just confirms what it found.
• Zero install friction — pure Python standard library. Nothing to pip install for it to work; if pip is missing it just runs from the clone.
• Stays alive across reboots — installs a cron launcher (@reboot + every minute) that nohups the dashboard and runs keepalive. Works headless over SSH, with no systemd --user / D-Bus / login-session requirement (the usual "Failed to connect to bus" trap).
• Restarts the right way — a tmux session can outlive a crashed agent; we detect that and relaunch the agent (e.g. claude --resume ), not just recreate an empty shell.

---

Quick start

curl -fsSL https://raw.githubusercontent.com/petrludwig-collab/AgentsMonitoring/main/install.sh | bash

…or from a clone:

git clone https://github.com/petrludwig-collab/AgentsMonitoring.git
cd AgentsMonitoring
python3 -m agentsmon setup

The wizard scans tmux, lists the agents it found, lets you pick which to supervise (proposing a restart command for each), optionally watches detected daemons, and installs the boot service.

To **spin up