Editorial Policy

How our guides, blog posts, and news articles get written, reviewed, and corrected — including where AI assists in drafting and what always requires a human decision.

Last updated: July 2026 · See our AI grading methodology for how the scanning/grading product itself works.

Donnie Laur

Founder, AgentGrail

1. How Content Is Made

AgentGrail publishes three kinds of content: guides (evergreen reference material — grading, identification, valuation), blog posts (timely market intel and set/release coverage), and news (time-sensitive coverage of new releases and events). Each type has its own section and URL path, and only genuinely timely content is published as news.

Articles are grounded in our own catalog data (checklists, print runs, market values) wherever possible, and in verifiable third-party sources (official manufacturer release pages, league schedules, established hobby publications) for facts outside our catalog. We don't invent product names, dates, or figures — if a fact can't be verified, it's left out rather than guessed.

2. AI-Assisted Content Disclosure

Some AgentGrail content is drafted with AI assistance — pulling structured facts from our catalog database, summarizing verified source material, and producing a first-pass draft. AI assistance is used for drafting speed, not as a substitute for verification: every fact used in a draft is checked against a real source (our own catalog or a cited external reference) before publish.

AI-assisted drafting never publishes on its own. Every article — whether written by hand or AI-drafted — goes through the same human review and publish gate described below before it goes live.

3. Human Review Before Publishing

Articles move through a Draft → Review → Published workflow. Publishing is a deliberate action, never automatic — a draft only becomes a live page when someone explicitly approves it.

Before that approval can happen, every article must clear a minimum content floor: enough real depth (not a thin stub), working cross-links to other real pages on the site (dead links block publish), and complete SEO/social metadata. This is enforced in code, not just editorial guidance — an under-floor draft is refused at the publish step with a specific list of what's missing.

Any HTML pulled in from an external source is also run through an allowlist sanitizer before it's stored, so a compromised or malicious source page can't inject executable content into a published article.

4. Partner and Syndicated Content

A small number of articles are republished from outside partners under an explicit redistribution agreement. Every partner-sourced article carries a visible attribution block linking back to the original author and source — we don't republish partner content without credit, and we don't republish anything marked as restricted from redistribution without confirmed permission first.

Partner content that references live prices or time-sensitive market calls is only published while that information is still current — stale market-timing content isn't left up past its useful window.

5. Corrections Policy

If you find an error in one of our articles — a wrong fact, a broken link, an outdated figure — we want to fix it. To report one:

  • Email [email protected] with the article URL and what's wrong
  • Use the feedback button available on the site

Corrections are made directly to the live article. For factual errors in our underlying catalog data (rather than article prose), see the feedback process on our methodology page.

Questions about how we produce content? Reach out — we're transparent about our process.

See our AI grading methodology →