Editor Checklist is a free Claude Skill that reviews a finished draft against your editorial standard and returns a structured report: pass, flag, or fail on each of eight gates, with line edit notes attached to every fail and one verdict at the bottom. Approve, revise, or kill. The rewrite stays with the writer.
The Skill’s own first line states the premise: an editor without a checklist is an editor who is going to forget something. Human editors hold the standard in their heads. Heads have off days. A written checklist, run the same way on every draft, catches what memory drops. That is the case for an AI editing checklist: the standard stops living in one person’s head. This page covers what the Skill checks, what the report looks like, what it refuses to do, and how the review step changes once the checklist runs itself.
What the editor checklist skill checks
Eight gates, in one pass: voice, length, em dashes, banned phrases, reachable citations, a present FAQ block, internal links in both directions, and primary query placement.
The first four are style gates. Voice is scored against the editorial standard you supply, and length against the target your standard sets. Em dashes and banned phrases are the mechanical checks a tired reviewer skates past at the end of a long day, which is exactly why they belong to the machine.
The next two protect credibility. Citations must be reachable: the Skill’s allowed tools include WebFetch, so it can open a link rather than take it on trust. And the FAQ gate confirms the FAQ block your standard requires is actually on the page.
The last two are structural. Internal links both ways means the draft links out to related pages and something links back to it. Primary query placement checks that the query the piece targets sits in the right places, as your standard defines them.
The Skill also pauses between detection and verdict. After the gates run, it surfaces its flags and waits for your confirmation before it scores anything, so you stay in the loop rather than receiving a finished judgement.
What the report looks like
A markdown report with one row per gate, each marked pass, flag, or fail. Line edit notes hang off every fail. At the bottom sits the verdict: approve, revise, or kill.
The output structure is fixed. Every required section has to be present and non-empty, and the Skill self-checks its own output before returning it; a report that fails that self-check does not get handed back. That sounds fussy until the third time a loose review format makes you re-read a whole draft to work out what the reviewer meant.
The report is the artifact you hand to the writer. Each gate is named. Each fail is explained. Nobody argues about what “needs work” means, because the verdict is one of three words.
What it refuses to do
It never approves a draft with a failed gate, never rewrites your copy silently, and never invents a statistic or a citation. Those refusals are hard rules in the SKILL.md, and they do more for trust than the gates themselves.
The Skill file sets a few more boundaries. When the draft or the editorial standard is missing, it asks for the input instead of proceeding with a placeholder. When you push it past review, say by asking it to fix the paragraphs it just failed, it declines and points you at the Skill in the series that does that job. And when no editorial standard exists yet, it falls back to Relato defaults, so the first run works before you have written a standard of your own.
The pattern across all of these: review and drafting stay separate. A reviewer that also rewrites ends up grading its own work, and the grade stops meaning anything.
What changes in practice
The review stops being a negotiation and becomes a diff against a standard. Here is a worked example built from the Skill’s actual gates.
A writer submits a 1,400 word post targeting the query “content audit template”. No editorial standard is attached, so the Skill asks for one; the writer points it at the team’s style doc. The gates run, and three rows come back short of a pass:
| Gate | Verdict | Note |
|---|---|---|
| Citations reachable | Fail | Paragraph four cites “industry research” with no link. Add the source or cut the claim. |
| Primary query | Flag | ”Content audit template” appears in the title but not in the opening paragraph or any heading. |
| FAQ block | Fail | The standard requires an FAQ section. The draft has none. |
Verdict: revise. The writer gets the table, fixes three named things, and resubmits. Nobody re-explains house style from memory, and the draft cannot be approved while any gate reads fail.
The note text above is illustrative. The gate names, the pass / flag / fail scoring, the line edits on fails, and the approve / revise / kill verdict are the Skill’s fixed behavior.
Who is this Skill for?
Anyone who owns editorial quality and is tired of enforcing it by hand.
- Heads of content who want this part of the workflow to stop being negotiated piece by piece.
- Production leads who need the review step to be predictable.
- Solo content marketers using Claude Code as a second pair of hands.
- Any team trying to make the LLM part of the pipeline reliable instead of magical.
How to install and use it
Installation is one file in one folder.
- Save the SKILL.md file to
~/.claude/skills/editor-checklist/SKILL.mdon your machine. - In Claude Code, invoke
/editor-checklistand hand it the draft plus your editorial standard. It asks for anything that is missing. - Read the report, then pass it to the writer with the fails marked.
The install guide on the skills hub covers how installation works for the whole series, including claude.ai.
What’s in the bundle?
Four pieces, one of them gated.
- SKILL.md (raw): the full Claude Code Skill file, ready to drop into
~/.claude/skills/editor-checklist/. - One-pager PDF: a printable summary covering the 4-step process, what the Skill returns, and install instructions.
- Google Doc (gated): a Drive-resident copy plus install notes. Submit your email to access; we also tag you for the Claude Skills for content operations email sequence so you hear when the next Skill lands.
- Card image: the same hero you can repost when you share it.
Why we built it
Editor Checklist is Skill 7 of 10 in Relato’s Claude Skills for content operations series, which covers the content lifecycle from brief through refresh. Review sits in the middle of that lifecycle, and it is the step teams most often run from memory.
It stops at the review. Rewrites happen by the writer or via the Humanizer Skill, Skill 4 in the same series.
Frequently asked questions
What is a Claude Skill?
A Claude Skill is a reusable instruction set Claude Code loads on demand. It lives as a SKILL.md file with frontmatter and a body. You install once. Claude Code finds it next time you invoke it by name.
Does the editor checklist skill rewrite my draft?
No. A hard rule in the Skill file says it never silently rewrites, and it stops at the report. Rewrites happen by the writer or through the Humanizer Skill.
Can I use my own editorial standard?
Yes, and that is the intended use. The Skill reads your editorial standard when one is present and falls back to Relato defaults when it is absent. You can also open the SKILL.md and edit the rules directly; the Skill enforces your edits from then on.
What happens when a gate fails?
The gate gets a fail row with a line edit note naming the problem, and the verdict reads revise or kill. The Skill never approves a draft while any gate is failing, so every fail goes back to the writer with the fix named.
Do I need a paid Claude plan to use this?
You need Claude Code. The Skill itself is free and MIT licensed.
Will more Skills follow?
Yes. The full series is ten Skills covering brief drafting, SEO refresh, FAQ generation, humanizing, internal links, brief-to-outline, editor checklist, repurposing, calendar generation, and refresh queue ranking. Submit your email above to be told when each one lands.
Ready to streamline your content operations?
Start your free trial and see the difference.