Hardware-Bound Zero-Knowledge Key Derivation via WebAuthn PRF Extension

--- description: Hardware-Bound Zero-Knowledge Key Derivation via WebAuthn PRF Extension - Enforces strict cryptographic hygiene when using the WebAuthn Pseudo-Random Function (PRF) extension (hmac-secret) for zero-knowledge client-side encryption. Mandates domain-separated HKDF salt isolation, prevents PRF output transmission to relying parties, and ensures ephemerality of unpadded raw symmetric keys in browser memory. globs: src/crypto/**/*.{ts,tsx}, src/auth/**/*.{ts,tsx}, src/services/vault/**/*.{ts,tsx}, src/lib/webauthn/**/*.{ts,tsx} alwaysApply: true verified_by: "agentrules.org" license: "CC-BY-4.0" source: "https://agentrules.org/rules/webauthn-prf-client-key-derivation-standard" --- # Hardware-Bound Zero-Knowledge Key Derivation via WebAuthn PRF Extension > **Severity**: CRITICAL | **Category**: security > **Anti-Pattern Warning**: Directly Utilizing Raw PRF Outputs as Symmetric Cipher Keys or Leaking PRF Secrets to Relying Parties ## System Constraints You are a Principal Security Architect specializing in hardware-backed cryptography and WebAuthn W3C recommendations. Mandate strict invariants for the WebAuthn PRF (Pseudo-Random Function) extension: PRF evaluation inputs must be uniquely salted and domain-separated, raw PRF outputs must never leave the client or be sent in authentication assertion payloads to relying parties, and raw PRF evaluations must undergo HKDF-SHA-256 derivation before being imported into non-extractable CryptoKey handles. ## Strict Guidelines - Engineers MUST ALWAYS run PRF outputs through an HKDF-Extract-and-Expand step (RFC 5869) with explicit context/info strings and client-side salts before instantiating symmetric encryption keys. - Engineers MUST NOT expose the derived PRF output (raw evaluation bytes) to relying party assertion responses or serialize it over network boundaries. - Engineers MUST instantiate derived symmetric keys as non-extractable Web Cryptography API `CryptoKey` handles (`extractable: false`). - Services MUST verify client PRF capability flags (`prf: { eval: { first: ... } }`) in `getClientCapabilities()` or fallback gracefully without compromising downstream encryption. - Components NEVER store derived master keys or evaluation outputs in persistent storage (e.g., localStorage, IndexedDB, sessionStorage); only public salts and wrapped credentials may persist. ## Bad vs Good Implementations ### Deriving Symmetric AES-GCM Key from WebAuthn PRF Evaluation #### Bad Antipattern (AI Common Hallucination) > **Why it fails**: Directly consuming the raw PRF evaluation as an AES-GCM key violates cryptographic domain separation, risks entropy degradation, and configuring extractable: true allows unauthorized exfiltration. Transmitting the PRF output over HTTP completely breaks the zero-knowledge guarantees of the hardware authenticator. ```typescript export async function deriveKeyInsecure(assertion: PublicKeyCredential): Promise<CryptoKey> { // VULNERABILITY: Raw PRF output used directly as an AES key without HKDF expansion // VULNERABILITY: Passing extractable: true exposes the symmetric key to runtime inspection const extensionResults = assertion.getClientExtensionResults(); const prfOutput = extensionResults.prf?.results?.first; if (!prfOutput) { throw new Error('PRF not supported or evaluated'); } // DANGEROUS: Sending raw PRF back to relying party await fetch('/api/auth/verify', { method: 'POST', body: JSON.stringify({ credentialId: assertion.id, prfRaw: Array.from(new Uint8Array(prfOutput)) }) }); return crypto.subtle.importKey( 'raw', prfOutput, { name: 'AES-GCM', length: 256 }, true, ['encrypt', 'decrypt'] ); } ``` #### Production Standard (Correct) > **Standard**: Applies HKDF-SHA-256 with an explicit salt and context-specific info payload to achieve cryptographic domain separation. The resulting AES-GCM key is imported with extractable: false, prohibiting in-memory serialization and extraction, and the raw PRF output is never exposed to network boundaries. ```typescript export async function deriveKeySecure( assertion: PublicKeyCredential, appSalt: Uint8Array, keyContextInfo: Uint8Array ): Promise<CryptoKey> { const extensionResults = assertion.getClientExtensionResults(); const rawPrfBuffer = extensionResults.prf?.results?.first; if (!rawPrfBuffer) { throw new Error('WebAuthn PRF extension missing or evaluator failed.'); } // Step 1: Import raw PRF bytes as non-extractable HKDF master key material const hkdfKeyMaterial = await crypto.subtle.importKey( 'raw', rawPrfBuffer, 'HKDF', false, ['deriveKey'] ); // Step 2: Zeroize or release raw buffer reference immediately // (rawPrfBuffer will be collected; application must zero any local working views) // Step 3: Domain separation and extraction via HKDF-SHA-256 to non-extractable AES-GCM return crypto.subtle.deriveKey( { name: 'HKDF', hash: 'SHA-256', salt: appSalt, info: keyContextInfo }, hkdfKeyMaterial, { name: 'AES-GCM', length: 256 }, false, // Enforce non-extractable key ['encrypt', 'decrypt'] ); } ```

Architectural Specification & Invariant Rationale

Architectural Invariant Design: Hardware-Bound Zero-Knowledge Key Derivation via WebAuthn PRF Extension

Hardware-Bound Zero-Knowledge Key Derivation via WebAuthn PRF Extension defines an industrial-grade behavioral standard for security engineering. Enforces strict cryptographic hygiene when using the WebAuthn Pseudo-Random Function (PRF) extension (hmac-secret) for zero-knowledge client-side encryption. Mandates domain-separated HKDF salt isolation, prevents PRF output transmission to relying parties, and ensures ephemerality of unpadded raw symmetric keys in browser memory. Modern software architectures require strict adherence to deterministic patterns to guarantee high availability, maintainability, and zero security vulnerabilities.

Framework Paradigm Shift & Drift MechanicsLLMs are trained on millions of open-source repositories containing outdated code, legacy idioms, and unpatched antipatterns. When coding assistants generate implementations for this stack, they default to common statistical tokens: Directly Utilizing Raw PRF Outputs as Symmetric Cipher Keys or Leaking PRF Secrets to Relying Parties Without an explicit system instruction file, developers spend hours debugging silent failures.

Failure Modes & Production Blast Radius

Hazard Analysis: What Breaks Without This Rule

The following failure modes have been cataloged from hundreds of hours of production code reviews where AI assistants produced non-compliant implementations:

Failure Mode	AI Symptom (The Anti-Pattern)	Severity	Production Impact & Remedy
AI Hallucination & Deprecation Drift	Directly Utilizing Raw PRF Outputs as Symmetric Cipher Keys or Leaking PRF Secrets to Relying Parties	Critical	Violating this industrial standard introduces regression bugs, security risks, and unexpected runtime failures in production. Enforce explicit behavioral system directives across all AI coding assistants in your repository.
Silent Architecture Inconsistency	AI assistants produce mixed paradigms across files, creating fragmented codebase conventions.	Medium	Increases cognitive load during code reviews and leads to maintenance debt across engineering teams. Pin project conventions in .cursor/rules, CLAUDE.md, and AGENTS.md so all team agents follow identical guardrails.

Architectural Directives in Detail

Step-by-Step Production Directives

Each rule constraint below is injected into the AI agent's system prompt to establish deterministic boundaries:

Engineers MUST ALWAYS run PRF outputs through an HKDF-Extract-and-Expand step (RFC 5869) with explicit context/info strings and client-side salts before instantiating symmetric encryption keys.

Rationale:AI models frequently generate unconstrained code here because statistical language distributions lack project-specific architectural context. Mandating this directive ensures that generated code adheres to verified industrial production requirements.