FleetGrader changelog

Changelog

What shipped, when.

Every entry below is a real tagged version — newest first. Some of the earliest ones predate the public launch; they're here because they're the reason everything after them works the way it does.

v1.1.3 — September 10, 2026

The MCP server no longer needs a source checkout — it's in the exe.

The same signed FleetGrader.exe you already download now runs its own MCP server directly — FleetGrader.exe --mcp enable then claude mcp add fleetgrader -- FleetGrader.exe --mcp-server, nothing else. No Python install, no private repo, no source access of any kind. Same four tools, same guarantee: an assistant gets an answer, never your client's real name.

v1.1.2 — September 9, 2026

Reports now collapse to fit the way you actually read them.

Every section of every report opens closed by default — click any one to expand it, click again to close it. A 14-section fleet report used to mean one long scroll; now it reads like a table of contents you can jump around in. Printing or exporting to PDF still shows everything, open or not, so nothing is lost, just tucked away until you want it.

v1.1.1 — September 9, 2026 — one dashboard, every client

Running fleets for more than one client? See all of them at once.

Fleet Rollup builds a single dashboard across every site you track — which issues are showing up where, health tiers site by site, pattern frequency across your whole book of business. It never shows one client's real hostname, serial, or display name to another; the aggregation is built from safe counts by construction, not scrubbed after the fact. Reachable from the app (Fleet Rollup, sidebar) or the CLI for scripting into your own reporting.

v1.1.1 — September 9, 2026 — drive it with an assistant

An AI assistant can check your fleet for you, without seeing your fleet.

FleetGrader ships a local MCP server: four tools an assistant like Claude can call directly, off by default until you turn it on. Every tool result is a strict subset of what the engine knows by construction — an assistant can tell you "one site has a new issue worth a look" without your client's real name, hostname, or serial ever landing in its transcript. Also this release: the first signed, publicly downloadable build since 1.0 — the public checkout was paused during a repositioning; this is the one that turned it back on.

v1.1.0 — September 2, 2026

Two new AppSpace report types: Visitor Directory and Visitor History.

Meeting-room and visitor-registration exports get their own report now, not a device-fleet score forced onto data that was never a device fleet. Visitor History surfaces no-shows and open visits — people checked in but never checked out, on the record still in the building. Visitor Directory flags incomplete records, duplicate emails, and registrations nobody has touched in a year.

v1.0.0 — August 31, 2026

1.0 — the first release you could actually buy.

Signed, gated, shipped: the first FleetGrader build a stranger could download, trial, and pay for. Everything before this tag was the engine finding its shape; everything after it is the product getting sanded down.

v0.6.1 — August 30, 2026

Every generated Excel report hardened against formula injection.

A crafted value in a vendor export — a device name, a content title, a filename — could otherwise plant a live formula that runs the moment the generated report opens in Excel. Every cell of every generated workbook is now sanitized before it's written, across every report type.

v0.6.0 — August 29, 2026

FleetGrader becomes an actual product: checkout, licensing, a signed installer.

$99/yr flat, live pricing, a real license key on activation, and a code-signed exe instead of a raw script. This is the release where "a tool I built" and "something I could sell" became the same sentence.

v0.5.0 — August 29, 2026

Korbyt Anywhere joins AppSpace and BrightSign.

FleetGrader's first platform beyond the original two: the same fleet-health scoring — status, platform mix, firmware currency, staleness — built against what Korbyt's own export actually supports, not a copy-paste of the AppSpace shape.

v0.4.0 — August 28, 2026

The desktop app arrives — FleetGrader becomes something you open and click.

Everything before this was engine-only, reachable by command line. This is the release with a window: drag-and-drop a report in, watch it build, browse your history, pick where reports land on disk.

v0.3.0 — August 28, 2026

FleetGrader starts remembering your clients — without mixing them up.

Site-keyed fleet memory: FleetGrader can now tell one integrator client's fleet apart from another's, so a pattern history stays with the site it belongs to. The foundation Fleet Rollup would later build its cross-client dashboard on top of.

v0.2.0 — August 28, 2026

The engine: report types, BrightSign log deciphering, the interactive report itself.

The jump from nothing to a working, tested, multi-format report tool across two platforms. This release is also where the BrightSign log decipherer's three-tier severity model and the self-contained, click-to-expand HTML report format — still the primary deliverable today — were born.

v0.0.1 — August 27, 2026

Day one.

The name, the domain, and the plan — nothing engine-side built yet. Everything above happened in the two weeks after this.

Questions

Wondering if something changed recently?

This page only lists what actually shipped. If you're not sure whether a fix landed yet, ask — I read every one of these myself.

[email protected]