Internal Link Auditor is a free Claude Skill that scans your blog folder and returns a link plan for one post: two to four outbound links from the new piece, one to three inbound links from existing posts, each with a file, an anchor, and a reason. It plans the links and leaves every file untouched.
What does the Internal Link Auditor skill do?
One linking pass for one post, in four steps.
- Step 1, point. You name one post, published or about to publish, plus the folder your blog lives in. The Skill asks for both explicitly. It will not run on placeholders, and when an input is missing it asks rather than guessing.
- Step 2, scan. It reads every other post in the folder and looks for three things: topical overlap with your piece, links that already exist, and candidate anchor phrases. Anything questionable surfaces as a flag, and the Skill waits for your confirmation before going further.
- Step 3, propose. Two to four outbound links from your post to related pieces, one to three inbound links from existing posts back to yours, with anchor text for every one.
- Step 4, hand over. The plan arrives as structured markdown in a fixed shape, every required section present and non-empty. Applying the links is a separate step, yours to take by hand or with a follow-up Skill.
What is an internal link audit?
An internal link audit checks how well one piece of content is wired into the rest of the site: which related pages it points to, and which pages point back at it.
The outbound half tends to take care of itself. Whoever wrote the post can link out on the day it ships, because the related pieces already exist and are fresh in mind. The inbound half is the part that gets skipped. The older posts that should point at the new one were written before it existed, and going back to edit them is a chore with no deadline. So the corpus drifts. Every publish adds a page, and the pages behind it stay silent about it.
This Skill scopes the audit to the width of the decision you are actually making. A site-wide link crawler hands you a report the size of your site; this runs when one piece ships and answers one question: which links should exist now that this post does.
What does the link plan look like?
Two sections of structured markdown, outbound from your post and inbound from existing posts, where every entry names the file, the anchor text, and the reason. Here is a worked sample for an invented post at /blog/content-calendar-template/:
Outbound from /blog/content-calendar-template/
1. /blog/content-audit/
anchor: "quarterly content audit"
reason: the planning section tells the reader to audit
before scheduling, and the phrase already sits in that
sentence
2. /blog/content-brief-template/
anchor: "a brief for each calendar slot"
reason: step two assumes a brief exists; this post
explains how to write one
Inbound to /blog/content-calendar-template/
1. /blog/content-operations-workflow/
anchor: "plan the quarter on a single calendar"
reason: the tooling section describes calendar planning
with no link to any calendar piece
The slugs, anchors, and reasons above are invented for the illustration; the shape is the Skill’s fixed output. Two rules show in the sample. Every anchor is a natural-language phrase already present in the host sentence, and every entry carries a reason a reviewer can veto. A plan you can veto line by line is a plan you can approve in one read.
What will it refuse to do?
Six refusals are written into the Skill file, and they carry most of its value.
- Editing anything. The output is a plan and the Skill stops there. Your files stay as they were until you, or a follow-up Skill, apply the links.
- Inventing URLs. Every target in the plan is a post the scan actually found in your folder.
- Manufacturing anchors. Anchor text always comes from natural-language phrases already in the text.
- Blowing past the caps. Four outbound and three inbound is the ceiling unless you ask for more.
- Touching exempt posts. A post marked exempt is skipped in the scan.
- Doing another Skill’s job. Pushed outside its scope, it declines and points you at the Skill in the series that covers the request.
Behind all six sits a self-check. The Skill verifies its own output before returning it and refuses to hand back a plan that fails. The file also bans invented statistics and citations outright, which matters less for a link plan than for prose, and still belongs in writing.
How do you install the Internal Link Auditor skill?
One file, one folder, one invocation.
- Save the SKILL.md file to
~/.claude/skills/internal-link-auditor/SKILL.mdon your machine. - In Claude Code, invoke
/internal-link-auditorand point it at one post and your blog folder. It asks for whatever you leave out. - Read the plan, then apply the links you approve, by hand or with a follow-up Skill.
Nothing runs on our servers and there is no account and there is no account, because the Skill is its instruction file. If you are installing several Skills from the series, the skills hub carries the general install guide once, rather than ten times.
Who is this Skill for?
People whose corpus grows faster than its links.
- A head of content whose older posts never mention the newer ones.
- An editor who places internal links from memory, on a blog that outgrew memory a hundred posts ago.
- A solo marketer shipping the blog from Claude Code, with no second person to run the linking pass.
- A team that wants the linking step to run the same way on every publish, whoever happens to publish.
What’s in the bundle?
Four files, one email gate.
- SKILL.md (raw): the complete Skill file, ready for
~/.claude/skills/internal-link-auditor/. - One-pager PDF: a printable page covering the four steps, the plan format, and the install path.
- Google Doc (gated): the same Skill as a Drive copy with install notes. The form on this page trades your email for the link and tags you for the Claude Skills for content operations sequence, and emails you as each one releases.
- Card image: the hero card, free to repost wherever you share the Skill.
Why we built it
Our own publish checklist on relato.com ends with an interlink pass: the new post links out to its cluster, and the existing posts that should point at it get updated to do so. The pass is dull, mechanical, and easy to skip, which is why ours lives in a written checklist rather than in anyone’s memory. Moving it into a Skill made it a step instead of an intention.
Internal Link Auditor is Skill 5 of 10 in Relato’s Claude Skills for content operations series, which runs from brief through refresh. Its nearest sibling is SEO Refresher, Skill 2, which updates a post’s internal links as one part of a full refresh. This Skill performs the linking pass alone, on a post you are not otherwise rewriting.
Frequently asked questions
What is a Claude Skill?
A markdown instruction file, SKILL.md, that Claude Code loads when you invoke it by name. Install it once. From then on it behaves like a command, same steps and same rules on every run, and everything it does is written in a file you can read before you trust it.
Does the Internal Link Auditor edit my posts?
No. It returns a plan and stops. A hard rule in the Skill file rules out silent edits, so your files are untouched when the run ends. Applying the links is a separate, deliberate step that stays with you.
How does it choose anchor text?
From phrases already in the text. The Skill reads the sentence where a link would live and picks a natural-language phrase out of it, which is why its anchors read like the author placed them. It never writes a new phrase into a post to carry a link.
Can it propose more than four links?
Yes, when you ask. The default caps, four outbound and three inbound, hold unless you request otherwise. We set them so a plan stays reviewable in one sitting; a reviewer who can hold the whole plan in view vetoes the weak entries instead of rubber-stamping the lot.
Do I need a paid Claude plan to use this?
You need Claude Code and nothing else. The Skill costs nothing and ships under an MIT license, so you can edit it, fork it, and hand it to your team.
Will more Skills follow?
Yes, and the next one is already live: Brief-to-Outline, Skill 6 of 10, turns a filled brief into an H2 and H3 outline before drafting starts. The full set of ten covers the lifecycle from brief drafting through refresh queue ranking, and the form above tags you for the series emails so each release reaches you when it lands.
Ready to streamline your content operations?
Start your free trial and see the difference.