ldlint

Finds schema.org mistakes that Google's Rich Results Test and most validators miss.

A fast CLI tool for validating JSON-LD against the schema.org vocabulary. Catches invalid properties, wrong types, and schema misuse that most tools don't detect.

Install

Prebuilt binaries

Download from the Releases page.

From source

go install github.com/kevhq/ldlint/cmd/ldlint@latest

Usage

# Validate HTML files in a directory (recursive)
ldlint dist/

# Validate specific files
ldlint index.html schema.json

# Validate a live URL
ldlint https://example.com/

# Pipe raw JSON-LD from stdin
echo '{"@type":"TouristTrip","guide":{}}' | ldlint -

# Mix and match
ldlint dist/ https://example.com/ -

# JSON output for CI
ldlint --format json dist/

# Refresh cached vocabulary
ldlint --refresh dist/

Example: catches an invalid guide property on TouristTrip

dist/en/tours/monte-alban/index.html
  TouristTrip
    ERROR  PROPERTY  "guide" is not valid on TouristTrip
                     ancestors: Trip > Intangible > Thing
                     valid on: CreativeWork, Event, ...
  LocalBusiness        OK (18 properties)
  WebSite              OK (5 properties)

1 file(s), 3 schema(s), 1 error(s)

Why not Google Rich Results Test?

Google's Rich Results Test only validates structured data for the specific rich result types it supports.

ldlint validates against the full schema.org vocabulary, including:

• Non-rich-result schemas (TouristTrip, Person, ItemList, etc.)
• Property misuse (valid property, used on the wrong type)
• Full type hierarchy validation with inheritance

Read more: Schema.org mistakes that Google's validator won't catch

Error Categories

• PROPERTY -- known type, invalid property (with "valid on" hint)
• SUPERSEDED -- warning: the term still works, but schema.org names a replacement (e.g. serviceArea -> areaServed)
• TYPE -- unknown schema.org type
• PARSE -- invalid JSON in a JSON-LD block. Reported against that block alone; the rest of the file is still validated
• EXTRACT -- the input itself could not be read (unreadable file, failed fetch, non-200 response)

Warnings are reported but never affect the exit code.

Exit Codes

• 0 -- no errors (warnings may still be present)
• 1 -- validation errors found
• 2 -- usage error, unreadable input path, or vocabulary fetch failure

How It Works

1. Fetches the schema.org vocabulary and caches it locally (~7 day TTL)
2. Builds a type hierarchy from rdfs:subClassOf and property domains from schema:domainIncludes
3. Extracts JSON-LD blocks from HTML using an HTML parser (not regex)
4. Validates each property against the type and all its ancestors
5. Flags terms marked schema:supersededBy and names their replacement

Flags

Flag	Description
--explain <Type>	Show valid properties and ancestors for a type
--format text|json	Output format (default: text)
--quiet	Only show schemas with errors or warnings
--verbose	Show property names for each schema
--no-color	Disable colored output
--refresh	Force refresh of cached vocabulary
--version	Print version and exit

Supported Input

• HTML files (extracts <script type="application/ld+json"> blocks, matched case-insensitively)
• Raw JSON-LD files (.json, .jsonld)
• URLs (http://, https://) -- fetched with a 30s timeout, redirects followed
• Stdin (-)
• @type as string or array of strings
• @graph containers and top-level arrays, unwrapped to one schema each -- from any source
• Single objects, arrays, or @graph wrappers