Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add tanghong123/claude-replay --skill monitor-fleetgit clone --depth 1 https://github.com/tanghong123/claude-replayWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/tanghong123/claude-replay/monitor-fleet)<a href="https://agentmods.dev/skills/tanghong123/claude-replay/monitor-fleet"><img src="https://agentmods.dev/badge/skills/tanghong123/claude-replay/monitor-fleet/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/tanghong123/claude-replay/monitor-fleet"><img src="https://agentmods.dev/badge/skills/tanghong123/claude-replay/monitor-fleet.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00084 | $0.00892 |
| Opus 5 | $0.00042 | $0.00446 |
| Sonnet 5 | $0.00017 | $0.00178 |
| Haiku 4.5 | $0.00008 | $0.00089 |
Grade A, and why
monitor-fleet scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 95 lines — stays where its author put it; the contents beside it link to each section on GitHub.
monitor-fleet
claude-monitor shows what is happening on one machine, over loopback only.
claude-monitor-fleet is a separate binary that opens one SSH tunnel per
environment and serves a switcher page with each machine's unmodified monitor
page inside it. The monitor itself is unchanged and needs no flag.
Discover before you configure
Never write a host name the user did not give you, and never assume a port. The tool ships with an empty config on purpose. Everything in it comes from discovery or from the user.
claude-monitor-fleet discover
This probes this machine and every literal Host in the user's own SSH config,
reads each monitor's port out of its lock file, and prints what exists. It writes
nothing. Show the user the list.
Then persist it:
claude-monitor-fleet discover --add
Or curate by hand when the user only wants some of it:
claude-monitor-fleet add prod --ssh prod-box
claude-monitor-fleet add laptop # no --ssh means this machine
claude-monitor-fleet remove prod
claude-monitor-fleet list
The config is JSON at $CLAUDE_MONITOR_FLEET_CONFIG, else
$XDG_CONFIG_HOME/claude-monitor/fleet.json, else
~/.config/claude-monitor/fleet.json — the user can edit it directly.
Serve the page
claude-monitor-fleet up
It brings up the tunnels, prints a http://127.0.0.1:<port>/ URL and opens a
browser. It stays in the foreground and owns the tunnels: they close when it
exits. So run it where it will outlive your turn — the user's own terminal is
the honest answer, and if you background it yourself, tell the user that closing
that shell takes the page down with it. --no-open skips the browser, --port N
pins the local port.
To look without saving anything to the config:
claude-monitor-fleet up --discover
When something is missing
Report what the tool reported. Do not fill gaps by guessing.
- No monitor on a host: it is not running there. Suggest starting
claude-monitoron that machine, oradd <name> --port Nif the user knows the port. - A host skipped during discovery: discovery uses
ssh BatchMode, so a host that would ask for a passphrase is skipped rather than left hanging. Suggest loading the key into an agent (ssh-add), or adding that host withadd, where prompts work. - Two monitors on one machine: they are found separately, each at its own published port, and told apart by cache root. That is expected, not a bug.
- An alias missing from the results: two
Hostentries for the same machine are collapsed to one environment — discovery says so on stderr when it happens. - A tunnel that dies later: the page's health dots go red and name the reason,
and the running
upre-opens that tunnel itself, with backoff. Report the reason and let it — restarting the fleet costs the tabs that were fine. A monitor that is down is a different thing: the fleet starts nothing, so that one stays down until someone starts it.claude-monitor-fleet statusre-probes from the command line.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 10d ago First seen · 95 lines · 84 tokens per session scan A 38de0a302134
monitor-fleet is a skill published in the GitHub repository tanghong123/claude-replay (2 stars, last pushed today), licensed MIT. It adds 84 tokens to every session and 892 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
ast-grep
Use for ast-grep project setup, rule authoring, rule debugging, and structural search workflows.
skill-installer
Install VT Code skills into /.agents/skills from a curated list or a GitHub repo path. Use when a user asks to list installable skills, install a curated skill, or install a skill from another repo, including private repos.
cmd-review
Review the current diff or selected files (usage: /review [--last-diff|--target |--file |files...] [--style ]).
cmd-analyze
Perform comprehensive codebase analysis and generate reports (usage: /analyze [full|security|performance]).
cmd-command
Run a terminal command (usage: /command [args...]).
skill-creator
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends VT Code's capabilities with specialized knowledge, workflows, or tool integrations.