Pre-Publish QA Checker

The problem this kills

You hit publish, and ten minutes later someone messages: the hero link is dead, the CTA points at last quarter's campaign, the lead image has no alt text, and there's a banned phrase legal asked you to stop using three times. Now you're editing live, the post already went out in the social queue, and trust in the content team takes a small hit.

The real problem isn't carelessness. It's that pre-publish checking is a manual, boring, easy-to-skip chore done under deadline. Humans are bad at scanning their own work for the same fifteen things every single time. Software is great at it.

This Implementation Plan gives you a tool that does the boring scan perfectly, every time, and then hands a clean report to a human to make the call.

What you'll build

A private web app for your content team where an editor:

1. Submits a draft - either pastes the HTML/markdown or gives a staging URL.
2. Watches automated checks run: every link followed (redirects included) and 404s flagged, every image checked for alt text, the text scanned for your banned phrases, every CTA checked for a proper UTM tag, and required elements (author, disclosure, CTA) confirmed present.
3. Reads a pass/fail report with line-level issues - exactly where each problem is.
4. Fixes and re-runs, or approves. Only an approved report flips the piece to "cleared to publish."

The rules - what's banned, what's required, how links must look - come from you, not from us. The build starts by interviewing you about your actual editorial standards.

What's inside the Implementation Plan

• A discovery interview as step one. Before a single line of code, the plan makes the AI agent interview you about your current publishing process, the tools and CMS you use, your exact required elements, your banned-word list, your link and UTM rules, your typical and peak draft volumes, and your messy edge cases. It reads back a short tailored spec and waits for your thumbs-up. You get a tool shaped to your standards, not a generic checklist.
• Step-by-step build instructions, each ending with a ready-to-paste prompt for your AI coding agent.
• A real data model for drafts, rule sets, check results, and approvals - shaped by your interview answers.
• The full link checker that follows redirects and flags dead links.
• The line-level pass/fail report and the human approval gate.
• A "No API yet?" fallback so you can run the whole thing today by pasting draft text and exporting the QA report as a CSV.

The governance it includes (this is the point)

This isn't a toy script. Governance is built in from the start:

• Login so only your team can use the tool.
• Row-level security so each organization only ever sees its own drafts and rules.
• A complete audit trail - who submitted, who approved, what changed, and when.
• A hard human-in-the-loop gate - the tool produces a report, but a person must approve before anything is marked cleared to publish. The AI drafts the verdict; the editor owns it.
• Duplicate guards keyed on draft URL/title plus version, so re-checks of the same piece are tracked as new runs, not silently duplicated.

Who it's for

Content ops managers, managing editors, and editors on teams that publish often and have been burned by a dead link or a wrong CTA going live. If "did anyone QA this?" is a recurring Slack message, this is for you. You do not need to be a developer - you need to know your own publishing standards.

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