Validate JWT Tokens Securely

This skill validates JSON Web Tokens - structure, cryptographic algorithms, security vulnerabilities, and validation best practices across languages and frameworks. JWT structure is a Header (algorithm and token type), Payload (registered, public, and private claims), and Signature (cryptographic integrity check). The critical validation steps run in sequence: verify the signature with the correct algorithm and key, validate the algorithm itself to prevent algorithm-confusion attacks (RS256 vs HS256), check the exp claim against current time with clock-skew tolerance, verify the iss claim matches the expected issuer, validate the aud claim contains the expected audience, and check nbf if present.

Use it when a service accepts JWTs from an external or multi-key issuer and needs algorithm-confusion protection and JWKS-based key rotation - not just a bare jwt.decode() call.

class JWTValidator:
    def get_public_key(self, kid):
        if kid in self.public_keys:
            return self.public_keys[kid]
        if self.jwks_url:
            response = requests.get(self.jwks_url)
            jwks = response.json()
            for key in jwks['keys']:
                if key['kid'] == kid:
                    public_key = jwt.algorithms.RSAAlgorithm.from_jwk(key)
                    self.public_keys[kid] = public_key
                    return public_key
        raise ValueError(f"Public key not found for kid: {kid}")

The Python JWTValidator class fetches a public key by kid from a JWKS endpoint (caching it), rejects any algorithm outside an explicit allowlist (RS256, RS384, RS512), then decodes with verify_signature, verify_exp, verify_nbf, verify_iat, verify_aud, and verify_iss all enabled plus require_exp/require_iat, layering custom claim checks (active-user status, required scopes as a set intersection) on top. Algorithm-confusion prevention is also shown in Go, asserting the parsed token's signing method is specifically RSA and specifically RS256 before trusting it, then verifying issuer and audience claims. A JavaScript token-inspection helper decodes without verifying signature, purely for debugging, and flags alg: "none" as critical, missing typ header, and computes time-to-expiry or flags an already-expired token. A production YAML config example sets allowed algorithms, issuer, multiple audiences, 30-second clock skew, required exp/iat, a JWKS URL with cache TTL and timeout, and custom validation for required scope and a 24-hour max token age.

Backend and API developers implementing JWT validation who need algorithm-confusion prevention and JWKS-based key rotation, not just a single hardcoded secret and default verify call. Eight key recommendations close the skill: never trust the algorithm from the token header without checking it against an allowlist, implement proper key rotation supporting multiple valid keys during rotation windows, use short access-token expiration (typically 15 minutes), validate every relevant claim rather than skipping issuer or audience checks, rate-limit token validation to prevent abuse, log validation failures to monitor for attacks, use established JWT libraries instead of a hand-rolled implementation, and cache public keys securely with proper invalidation.