Facility Safety Walkthrough & Findings
Build a mobile internal tool for periodic safety walkthroughs: log hazards with photos from your phone, have a supervisor approve severity and owner, and track every corrective action to closed - with aging and a clean CSV export.
A phone-friendly, login-protected walkthrough tool: run an inspection on your phone, log findings with photos, have a supervisor set severity and approve the corrective action + owner, track every open finding to closed with aging, and export findings as CSV - with a hard rule that nothing becomes a tracked corrective action until a supervisor approves it.
Before you start
- A free Vercel account
- A free Supabase account
- A free Resend account (and a sender address you can use)
- A walkthrough checklist template (the items you inspect)
- A location/zone list and an action-owners list (CSV or Google Sheet)
The problem this kills
Your safety walkthrough lives on a clipboard, in a phone's camera roll, and in three half-finished email threads. Someone spots a blocked fire exit on Tuesday, snaps a photo, means to send it to maintenance, and forgets. The same trip hazard by the loading dock gets "found" on every walkthrough for a month because nobody opened a real action against it. When an auditor or an insurer asks "show me the open safety findings and when you closed them," you're piecing it together from memory.
The walkthrough itself isn't the problem - your team knows what to look for. The problem is what happens after the finding: it has no owner, no due date, no severity, and no record that anyone followed up. Photos sit on a phone. Repeat hazards never get flagged as repeats. And "we fixed it" is a verbal claim, not a tracked, dated, closed item.
This tool turns each finding into a tracked corrective action with an owner, a due date, and a paper trail - and it puts the whole walkthrough on your phone.
What you'll build
A small internal web app, just for your team, that:
- Lets an inspector run a walkthrough on their phone against your checklist, zone by zone.
- Logs each finding with photos captured straight to secure Storage - blocked exit, trip hazard, spill, missing signage, whatever your list covers.
- Sends each finding to a supervisor, who reviews it, sets the severity, and approves the corrective action, owner, and due date before it becomes a real open item.
- Turns approved findings into tracked corrective actions with an owner and deadline, and follows them all the way to closed.
- Shows an open-findings report with aging - what's overdue, what's coming due, by zone and severity.
- Dedupes repeat hazards - if the same kind of hazard keeps showing up in the same zone, it flags it as a repeat instead of quietly opening a fifth identical action.
- Supports a recurring walkthrough schedule so inspections actually happen on time.
- Exports findings as CSV for your records, your safety committee, or whatever system you report into.
What's inside the Implementation Plan
The plan is a single markdown file you paste into Claude Code (a free AI coding agent). It walks the agent through building the whole tool, step by step, each step ending with a ready-to-paste prompt.
The most important part: the plan opens by interviewing you about your facility. Before it writes a single line, the agent asks what your walkthrough checklist actually contains, how your zones and locations are named, who your action owners are, what your severity levels mean, and what your messiest cases look like (a hazard that spans two zones, an owner who left, a finding that's an immediate stop-work). It reads a short tailored spec back to you, you confirm it, and only then does it build - so you get a tool shaped to your site and your checklist, not a generic template you have to fight.
Inside you'll find:
- The discovery interview and how the agent turns your answers into the data model.
- The full build: database, login, the mobile walkthrough flow, photo capture to Storage, the supervisor review-and-approve gate, corrective-action tracking with aging, repeat-hazard dedupe, and the recurring schedule.
- The hard human-in-the-loop approval gate so no finding becomes a tracked action without a supervisor's sign-off.
- Verification steps so you can prove it works, and the CSV-export fallback so it's fully usable even before you connect it to any other system.
The governance it includes (this is the point)
This isn't a toy. The plan builds in the controls a safety program actually needs:
- Login so only your team can run walkthroughs or see findings.
- Row-level security so people only see their own organization's (or site's) data.
- A complete audit trail - every finding, photo, severity change, approval, reassignment, and closure is logged with who and when.
- A hard human-in-the-loop gate - the inspector logs a raw finding, but a supervisor must review and approve severity, owner, and due date before it becomes an open corrective action. Nothing auto-opens.
- Duplicate guards so the same hazard in the same zone can't quietly become five separate open items.
Who it's for
Safety officers, facilities supervisors, EHS coordinators, and store/site managers who run periodic walkthroughs and are tired of findings dying in a camera roll. If you want a real, auditable corrective-action tracker without buying a heavyweight EHS platform or hiring a developer, this is for you. You don't need to write code. You need your checklist, your zone list, your owners list, and an afternoon.
You've got this - paste the first prompt and let the agent interview you.