agentsclimarketplace

Retail pos fleet health

Skill whobat/AI-Agent-skills/skills/ax-retail/retail-pos-fleet-health

a collection of AI Agent skills

Install
npx -y skills add whobat/AI-Agent-skills --skill retail-pos-fleet-health

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Health sweep across a fleet of Windows POS machines (Dynamics AX 2012 R3 Retail POS or similar) — or any Windows server/workstation fleet — over PowerShell Remoting. The bundled script checks each machine in parallel for stopped auto-start services (POS/Retail/SQL by default), local SQL Server instances with per-database sizes (flagging databases approaching the SQL Server Express 10 GB limit), low disk space, and recent Critical/Error event counts, returning ranked warnings as JSON. Use when the user wants to check/sweep/inspect the POS fleet or a set of Windows machines — e.g. "check all the POS machines", "which tills have problems", "is any POS offline database near the Express limit", "sweep the store machines for stopped services". Read-only. Requires PowerShell 7+, WinRM on the targets, and an admin credential (always prompted).

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

4.4 KB, 882 tokens by cl100k_base, as published. Nobody here has run it

Retail POS Fleet Health

Sweeps many Windows machines in parallel over WinRM. The bundled script scripts/Invoke-PosFleetHealth.ps1 collects health signals and emits JSON with deterministically ranked warnings; the agent (you) writes the narrative. The script is read-only on the targets — it never starts/stops anything.

SCRIPT = this skill's scripts/Invoke-PosFleetHealth.ps1.

Credentials

The script always prompts for an admin credential (Get-Credential) — held in memory for the run, reused across all hosts, never written to disk. Do not pass passwords on the command line. SQL checks run inside each remote session against localhost with that Windows login — no SQL credentials needed.

How to run

Always run with pwsh. Parse the JSON from stdout.

WantPass
A few machines-ComputerName POS001,POS002,POS003
The whole fleet-ServerListFile C:\ops\pos-hosts.txt (one host per line, # comments OK)
Other services-ServicePattern '*POS*','*Retail*','MSSQL*','MyService*'
Stricter disk check-DiskMinFreePct 15
Event window-EventHours 24 (default)
Express limit warn level-ExpressWarnAtPct 80 (default; of -ExpressLimitGB 10)
Parallelism-ThrottleLimit 12 (default)
Save full report-OutFile C:\ops\fleet-health.json (do this for 100 hosts!)
Transport/auth-UseSSL · `-Authentication Negotiate
# Full fleet sweep, detail to file, compact summary on stdout
pwsh -File SCRIPT -ServerListFile C:\ops\pos-hosts.txt -OutFile C:\ops\fleet-health.json

Output contract

  • Without -OutFile → full JSON (all hosts) on stdout.
  • With -OutFile → full detail to the file; stdout gets a compact summary (status, host counts, ranked warnings). For big fleets always use -OutFile.

summary.warnings is ranked: unreachable/auth → stopped services → Express DBs near the limit → low disk → SQL query failures → high error counts. Each warning has computer, type, detail. Per-host detail (disks, services, SQL instances + database sizes, recent error counts) is under hosts[]. See REFERENCE.md.

What you (the agent) do with the result

  1. Run the script, parse the JSON.
  2. Lead with summary.warnings — grouped by type, named by machine. A fleet sweep answer is "which machines need attention and why", not 100 host reports.
  3. Always surface coverage: report hosts_failed (unreachable/auth) explicitly — an unreachable POS may itself be the incident. Never imply full coverage when hosts failed.
  4. Express limit warnings are urgent: a POS offline database hitting the 10 GB data-file limit stops syncing/working. Recommend cleanup/shrink or escalation per the organization's POS procedure.
  5. For deeper SQL diagnosis on a flagged machine, follow up with sqlserver-perf-triage (or ax2012-sql-performance for the channel DB); for event-log detail, use win-eventlog-triage against that host.

Errors

  • Many unreachable hosts → WinRM not enabled on POS images, firewall, or wrong network.
  • auth_failed → the credential lacks admin rights on the targets.
  • sql_query_failed on a host → SQL service running but the login has no access; the rest of the host's data is still collected.

What ships with it: 4 files

27.9 KB alongside SKILL.md, 2 of them executable

scripts/

Keep looking

Skills are one crate of 327,069. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.