§ 4@integrity & context allowlist

Remote context loading is the largest attack surface in JSON-LD. The security extensions add Subresource-Integrity-style hashes and a first-class allowlist so a document specifies — verifiably — which contexts it expects.

Input

algorithm

context (canonical)

tampered version (e.g. attacker-replaced)

Output

integrity hash

computing…

original

✓ verified

tampered

✗ blocked

Context allowlist

A second layer: reject any context URL that is not in an explicit allowlist. Patterns support a single trailing * wildcard.

url to test

allowed (one per line)

patterns (one per line)

blockRemoteContexts

{
  "url": "https://schema.org/",
  "config": {
    "allowed": [
      "https://schema.org/",
      "https://www.w3.org/ns/"
    ],
    "patterns": [
      "https://w3id.org/*"
    ],
    "blockRemoteContexts": false
  },
  "allowed": true
}