HSM / PKCS#11 / Multi-Algorithm Signing
What you will do: Choose a JWT signing algorithm and sign with the browser HSM. Verify the original payload with the local public key, change the received subject, then restore it.
Worked example: With ML-DSA-65, the signature grows from the 256-byte RS256 baseline to 3,309 bytes. Verification accepts the original payload, rejects the changed subject and succeeds when the original value is restored.
Runtime and privacy: The cryptographic exercise runs in this browser. Review the site privacy terms before entering sensitive material; use synthetic inputs for learning and evaluation.
Browse all Crypto Lab tools · Learn with Identity & Access Management with PQC
For your role
- Developer / Engineer
- Switch the signing algorithm from RS256 (RSA-2048) to an ML-DSA variant, press Sign Token (Live WASM) and then Verify Signature: the header, the signature size and the Size Impact Analysis show what your JWT libraries and token limits will see.
- Security Architect
- Use the Size Impact Analysis and the CBOR Token Encoding (CWT / COSE) alternative to decide whether a PQC token still fits your headers, cookies and gateways before choosing the algorithm.
- Researcher / Academic
- Run the IAM PQC Known Answer Tests panel, then compare the same payload signed with each algorithm to measure the exact signature and header overhead.
- IT Ops / DevOps
- The Verification Flow — Relying Party Perspective panel shows what every service that validates tokens must be able to parse after the switch; use it to list which relying parties need updating first.
Token Signing Migration Lab
Compare JWT signing algorithms side-by-side. Toggle between classical RS256/ES256 and quantum-safe ML-DSA variants to observe signature size impact and header changes.
Initializing SoftHSM…
C_Initialize → C_InitToken → C_OpenSession → C_Login
Select algorithm to see header changes and signature size impact
{
"alg": "RS256",
"typ": "JWT",
"kid": "rsa-2048-key-001"
}alg field must match the signing key type. JWK Set endpoint must expose the corresponding PQC public key.{
"sub": "user-1234",
"name": "Alice Engineer",
"email": "alice@corp.example",
"roles": ["developer", "ci-pipeline"],
"iat": 1741392000,
"exp": 1741395600,
"iss": "https://auth.corp.example",
"aud": "https://api.corp.example"
}Size Impact Analysis
Encoded header 87 + payload 324 + signature 342 + 2 separators = 755 ASCII bytes. This sizing calculation uses the displayed signature byte count; it does not establish a valid, interoperable token or validate the signature encoding. HTTP and cookie framing are excluded.
- • Compare the compact token plus Authorization/Bearer framing and other request headers with the limits documented for the selected server and configuration
- • For cookie transport, account for the cookie name, attributes and the selected browser/server limits. Token length alone does not establish fit
- • Load balancer JWT validation: Envoy, Kong, AWS ALB must update JWT algorithms allow-list to include ML-DSA variants
- • Client SDK updates: JOSE/JWT libraries (jose, jsonwebtoken, PyJWT) need PQC algorithm support — check vendor roadmaps
Alternative: CBOR Token Encoding (CWT / COSE)
JWT uses text-based JSON encoding. For constrained environments (IoT, mobile credentials, embedded systems), CBOR Web Tokens (CWT) with COSE signatures offer a binary-encoded alternative that reduces overhead — important when ML-DSA signatures already add kilobytes.
- • Text (base64url JSON)
- • Wide browser/library support
- • RFC 9964 (ML-DSA for JOSE and COSE)
- • Published May 2026
- • Binary (CBOR-encoded)
- • Smaller envelope overhead
- • RFC 9964 (ML-DSA for JOSE and COSE)
- • Used in ISO 18013-5 mDL
- • IoT / embedded: CWT/COSE
- • Mobile credentials: CWT/COSE
- • Enterprise SSO / web: JWT
- • Both: same ML-DSA signature
Both JWT and CWT use the same underlying ML-DSA (FIPS 204) signature — only the token envelope encoding differs. See RFC 9964 and RFC 9052 in the Library for specification details.
Verification Flow — Relying Party Perspective
After the IdP signs a JWT, the relying party must verify it. The verification chain: fetch JWKS endpoint → match kid → verify signature with the public key. Enable Live WASM and sign a token first to run live verification.
{
"keys": [{
"kty": "RSA",
"alg": "RS256",
"kid": "rsa-2048-key-001",
"n": "sLjA...long modulus...",
"e": "AQAB",
"use": "sig"
}]
}GET /.well-known/jwks.json — resolve public key matching kid in JWT header
base64url(header) + "." + base64url(the received payload)
PKCS#11 v3.2 verification with the local RS256 public key used in this exercise
Enable Live WASM above, then sign a token to unlock verification.
Note: In illustration mode, signature data is representative. Enable Live HSM mode above to execute real PKCS#11 v3.2 signing and verification operations across all five algorithms. Keys generated here are for educational use only — in production IAM deployments, verify your HSM/KMS active FIPS 140-3 CMVP certificate, IdP/JWKS RFC 9964 support, and proxy/gateway HTTP header limits.
IAM PQC Known Answer Tests
OpenID Connect · SAML 2.0 · FIPS 204 · FIPS 198-1 · NIST SP 800-132
Predict a one-byte change
This fictional sample generates fresh ML-DSA-65 keys. Predict whether changing the selected byte should preserve verification or key confirmation, then run the actual operation.
Click Run known-answer checks to run 5 use-case scenarios. Evidence in this set: NIST ACVP-Server reference sample — Expected values copied from the public NIST ACVP-Server repository with immutable source identity; Published standard KAT — Expected values printed in a cited standard or consensus RFC; Functional round-trip — Output produced by an implementation is consumed by the same or paired implementation.
Reference samples from the public NIST ACVP-Server repository · OpenID Connect · SAML 2.0 · FIPS 204 · FIPS 198-1 · NIST SP 800-132 · Generated keys are for educational use only.
Known issue: the password-derivation (PBKDF2) test errors with CKR_ARGUMENTS_BAD — this is a softhsm-wasm C_DeriveKey binding limitation, not a cryptographic failure. The other KAT rows are unaffected.
Production Tool Capabilities & Verification Requirements
What this sample check establishes: Demonstrates cryptographic token and PKCS#11 partition migration steps from legacy classical mechanisms (`CKM_RSA_PKCS`, `CKM_ECDSA`) to PKCS#11 v3.2 PQC mechanisms.
Browser sandbox limitation (non-certification boundary): Uses an in-browser softHSM emulation; it does not test vendor-specific HSM cloning protocols, M-of-N smartcard quorum ceremonies, or PCI key blocks.
What to require and test in your production tools:
- OASIS PKCS#11 v3.2 compliant HSM client libraries supporting `C_EncapsulateKey`, `C_DecapsulateKey`, `CKM_ML_KEM`, and `CKM_ML_DSA`.
- Dual-stack partition policies allowing co-existence of legacy classical keys and new PQC keys during phased application cutover.
- Non-exportable private-key attributes (`CKA_SENSITIVE=TRUE`, `CKA_EXTRACTABLE=FALSE`) verified by HSM audit logs.
Try it
- 1. Predict before you run it.
- 2. Run or change the tool above.
- 3. Explain what the result shows.
Sign a token with RS256 and then with an ML-DSA variant. What in the Size Impact Analysis grows most?
Next step
Turn it into a plan: Hybrid Transition PlannerThis tool practises the Identity & Access Management with PQC module, phase 5 (Pilots & Migration); Hybrid Transition Planner produces a deliverable of that phase.
Related content
Next in HSM / PKCS#11