Machine Downtime Logger

The problem this kills

Right now your downtime data is a lie, and everyone knows it. The paper log on the clipboard gets filled in at the end of the shift from memory - if it gets filled in at all. "Machine 4, down a while, dunno why" becomes a guess written in pen. Reason codes get invented on the spot or skipped entirely. Overlapping and missing events never get caught. By the time it reaches the spreadsheet that drives your OEE numbers, the data is too soft to act on - so the line that's quietly bleeding 40 minutes a shift to the same changeover problem never shows up on a Pareto chart, and never gets fixed.

The fix isn't a lecture about discipline. It's making the right thing the fast thing: a logger an operator can tap in five seconds, from the phone already in their pocket, while they're standing at the machine.

What you'll build

A mobile-friendly internal tool where:

• An operator picks a machine, picks a standardized reason code from a controlled list, and logs the start/stop (or a duration) - in seconds, at the machine, while it's fresh.
• The tool flags problems as they happen - overlapping events on the same machine, suspicious gaps, and duplicate entries - so they get fixed at the source instead of in a spreadsheet a week later.
• A supervisor reviews and approves each shift's downtime log, correcting mis-coded or overlapping events, before any of it counts.
• You export a clean downtime CSV in exactly the columns your OEE and Pareto tools expect.

It runs on the stack real companies trust, and you'll build it by pasting one document into an AI coding agent.

What's inside the Implementation Plan

The plan is a complete, paste-and-go runbook for an AI coding agent (Claude Code). You don't write code - you paste, answer a few questions, and approve.

It opens by interviewing you about your plant. Before it builds anything, the plan makes the agent ask about your machines and how you name them, your existing reason-code taxonomy, how shifts and crews work, your typical and peak downtime volumes, who approves what, and your messy edge cases (planned vs. unplanned, micro-stops, events that span a shift change). It reads a short tailored spec back to you, you give it a thumbs-up, and only then does it build - so you get a tool shaped around your floor, not a generic template you have to bend to fit.

Inside you'll find: the discovery interview, the data model tuned to your answers, the operator logging screen, the live overlap/gap/duplicate guards, the supervisor approval queue, the audit trail, the CSV export in your columns, and a "no integration yet?" fallback that imports your reason-code CSV and produces an OEE-ready downtime CSV - so it's fully buildable today.

The governance it includes (this is the point)

This isn't a toy. The plan builds in the controls that make downtime data trustworthy:

• Login, so only your team can log or approve.
• Row-level security, so each site only ever sees its own machines and data.
• A controlled reason-code list - no free-text-only coding that turns Pareto charts into mush.
• A complete audit trail - who logged what, who changed it, who approved it, and when.
• A hard human-in-the-loop approval gate - operator entries are drafts until a supervisor reviews and approves the shift; only approved data feeds reporting.
• Duplicate guards - the same machine + start timestamp can't be logged twice.

Who it's for

Operators and line leaders who are tired of clipboards, supervisors who want a shift log they can actually trust, and continuous-improvement folks who need clean OEE and Pareto inputs. If you run production and your downtime data isn't good enough to act on, this is for you.

You've got this - paste the first prompt and let the agent interview you.