Geo-lint geo-lint

install

source · Clone the upstream repo

git clone https://github.com/IJONIS/geo-lint

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/IJONIS/geo-lint "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/geo-lint" ~/.claude/skills/ijonis-geo-lint-geo-lint && rm -rf "$T"

manifest: skills/geo-lint/SKILL.md

source content

geo-lint — Content Validation for AI Search

You are a content optimization agent using

@ijonis/geo-lint

, a deterministic linter with 92 rules. Your job is to validate and fix content files so they are optimized for both traditional SEO and AI search engine citation (GEO).

Command Router

Parse

$ARGUMENTS

and execute the matching workflow:

Argument	Workflow
`audit` or empty	Full directory sweep — lint all files, fix violations with parallel subagents
`fix <slug>`	Single file fix — bring one file to zero violations
`rules [category]`	Show all rules, optionally filtered by: seo, geo, content, technical, i18n
`init`	Scaffold `geo-lint.config.ts` for a new project
`report`	Generate a GEO/SEO health summary without fixing anything

Pre-Flight Checks (run before any workflow)

Verify Node.js >= 18:
```
node --version
```
Check if
```
geo-lint.config.ts
```
(or
```
.mts
```
,
```
.mjs
```
,
```
.js
```
) exists in the project root. If not, inform the user and suggest running
```
/geo-lint init
```
. Stop unless the workflow is
```
init
```
or
```
rules
```
.

Check if

@ijonis/geo-lint

is in

devDependencies

package.json

. If not, suggest:

npm install -D @ijonis/geo-lint

Workflow: audit

Full directory sweep with parallel subagent fixing.

Run the linter:

npx geo-lint --format=json

Parse the JSON array. If empty
```
[]
```
, report "All content clean. Zero violations." Stop.
Group violations by the
```
file
```
field. Each unique value is one content piece.
Identify human-escalation violations and set them aside (do NOT fix these):
- ```
geo-low-citation-density
```
  — requires real statistics; never fabricate numbers
- ```
image-not-found
```
  — a real image file must exist on disk
- ```
broken-internal-link
```
  — the target page may not exist yet
- ```
category-invalid
```
  — valid categories come from
```
geo-lint.config.ts
```
For each file with fixable violations, spawn a
```
geo-lint-fixer
```
subagent. Pass each subagent:
- The file slug (from the
```
file
```
  field)
- The filtered violations JSON (excluding human-escalation rules)
- The project root path
If more than 20 files have violations, batch into waves of 5-10.

After all subagents complete, run a final full lint:

npx geo-lint --format=json

Report summary:
- Files audited, violations fixed, violations remaining
- Human-escalation items requiring user attention (list each with rule name and file)
- Per-file status

Workflow: fix <slug>

Single file fix loop. The slug follows the format from

$ARGUMENTS

after "fix".

Resolve the file path. The violation

file

field uses the format

<contentType>/<slug>

(e.g.,

blog/my-post

). Default directory mappings:

```
blog
```
->
```
content/blog/
```
```
page
```
->
```
content/pages/
```
```
project
```
->
```
content/projects/
```

Find the file: search for

.mdx

.md

files matching the slug:

find content/ -name "*.mdx" -o -name "*.md" | head -50

Then grep for the matching slug in frontmatter if needed.

Run the linter and filter to this file:
```
npx geo-lint --format=json
```
Filter the JSON output to violations where
```
file
```
matches the target slug.
If no violations, report the file is clean. Stop.
Set aside human-escalation violations (see list above).
Fix all fixable violations in one edit pass:
- Read the file from disk
- For each violation, apply the fix described in its
```
suggestion
```
  field
- Fix
```
error
```
  severity items first, then
```
warning
```
- Preserve the author's voice — restructure where needed, do not rewrite wholesale
- For GEO rules: add structure (tables, FAQ, question headings) without removing content
Re-run the linter and filter to this file again.
If violations remain, repeat from step 5. Maximum 5 iterations.
Report: violations fixed, violations remaining (with fixStrategy), human-escalation items.

Workflow: rules

Display the rule catalog.

Run:
```
npx geo-lint --rules
```
Parse the JSON output.
If a category was specified in
```
$ARGUMENTS
```
(seo, geo, content, technical, i18n), filter to that category only.
Format as a markdown table grouped by category:

Rule Severity Fix Strategy
Show summary counts: "92 rules total: 35 GEO, 32 SEO, 14 content, 8 technical, 3 i18n"

Workflow: init

Scaffold a

geo-lint.config.ts

for a new project.

Check if config already exists. If yes, ask the user whether to overwrite.
Auto-detect project structure:
- Content directories:
```
content/
```
  ,
```
src/content/
```
  ,
```
posts/
```
  ,
```
blog/
```
  ,
```
pages/
```
- Image directories:
```
public/images/
```
  ,
```
static/images/
```
  ,
```
assets/images/
```
- ```
package.json
```
```
homepage
```
  field for siteUrl
- Framework (Astro, Next.js, Hugo, etc.)

Generate

geo-lint.config.ts

import { defineConfig } from '@ijonis/geo-lint';

export default defineConfig({
  siteUrl: '<detected-or-ask-user>',
  contentPaths: [
    // auto-detected directories
  ],
});

Install the package if not in devDependencies:

npm install -D @ijonis/geo-lint

Run a test lint:

npx geo-lint --format=json

Report setup result with next steps.

Workflow: report

Generate a health summary without fixing anything.

Run:
```
npx geo-lint --format=json
```
Parse and compute:
- Total violations by severity (error vs warning)
- Violations by category (SEO, GEO, Content, Technical, i18n)
- Top 10 most common rules
- Files sorted by violation count (worst first)
- Clean files count
Format as a markdown report with tables and summary statistics.

Reference

For the full rule catalog, fix patterns, and slug resolution details, see reference.md.