Site / Supplier Audit Scorecard & Report Generator

The problem this kills

You audit sites or suppliers against a standard, and the scoring lives in a spreadsheet that someone built once and everyone has since "improved" into five incompatible versions. Weights get fat-fingered. A critical finding that should auto-fail the whole audit gets buried under a healthy-looking 87%. Two auditors score the same checklist two different ways. And the report? It's copy-pasted from the last one, half the action items get lost, and nobody can tell you whether this site is trending up or down across its last three visits.

The math is not the hard part. The hard part is doing it the same way every time, capping the score when a critical question fails, keeping the evidence attached, making one person sign off before anything goes out, and producing a report that looks like your report - not a generic template.

What you'll build

An internal web app, just for your team, that turns a scoring template into a clean, repeatable audit:

• Load your template - sections and questions, each with points/weights and a pass threshold, including critical (auto-fail) questions that cap the result no matter how good the rest looks.
• Score an audit - the auditor answers each question, adds notes, and attaches photo evidence as they go.
• Automatic scoring - the tool computes per-section scores and an overall weighted result, applies any auto-fail caps, and places the audit in a Pass / Conditional / Fail band.
• A drafted report narrative - the agent writes a first-draft report (summary, per-section findings, and action items) from the answers.
• A human approval gate - the lead auditor reviews the computed score and the draft, edits the narrative and action items, and only then approves. Approval is what issues the report.
• Issue + export - the approved report is emailed to the site via Resend, the action items are released, and you get a scorecard CSV (in your column layout) and a PDF.
• Trend - each site's score is tracked across audits so you can see whether it's improving.

What's inside the Implementation Plan

• A copy-paste runbook you feed to an AI coding agent (Claude Code) - no prior coding needed.
• It starts by interviewing you about your business. Before it builds anything, the plan makes the agent ask about your real scoring template, your weighting and pass rules, how you handle critical/auto-fail questions, your site and supplier naming, your evidence and report conventions, and your messy exceptions. It reads a short tailored spec back to you and waits for your thumbs-up. You get a tool shaped to your standard, not a generic checklist app.
• Step-by-step build instructions, each ending in a ready-to-paste prompt.
• The full data model for templates, audits, answers, evidence, and reports - tuned to your answers.
• A "No API yet?" fallback: import your existing scoring spreadsheet/CSV as the template and export a scorecard CSV in the exact columns your current system expects, so it's fully usable today even with zero integration.

The governance it includes (this is the point)

This isn't a toy. The plan builds in the controls an audit function actually needs:

• Login so only your team can use it.
• Row-level security so each auditor and organization sees only their own audits and sites.
• A complete audit trail - who scored what, who edited the draft, who approved, and when.
• A hard human-in-the-loop approval gate - the AI drafts the report and the score, but nothing is issued and no action items are released until the lead auditor reviews and approves.
• Duplicate guards - the same site on the same audit date can't be scored twice (dedupe key = site + audit-date).

Who it's for

Supplier-quality engineers, EHS auditors, and franchise / operations auditors who score sites against a standard and have to produce a defensible, consistent report every time.

You've got this. Open the Implementation Plan, paste the first prompt, and answer the interview - the tool builds itself around how you already audit.