Claude-Red

SKILL: OAuth Security Testing

Metadata

• Skill Name: oauth-attacks
• Folder: offensive-oauth
• Source: https://github.com/SnailSploit/offensive-checklist/blob/main/oauth.md

Description

OAuth 2.0 attack checklist: authorization code interception, redirect_uri bypass, CSRF on OAuth flow, state parameter abuse, open redirector chaining, token leakage via Referer, PKCE bypass, and scope escalation. Use when testing OAuth implementations in web apps or bug bounty.

Trigger Phrases

Use this skill when the conversation involves any of: OAuth, OAuth 2.0, authorization code, redirect_uri bypass, OAuth CSRF, state parameter, PKCE bypass, scope escalation, token leakage, open redirector, OAuth attack

Instructions for Claude

When this skill is active:

1. Load and apply the full methodology below as your operational checklist
2. Follow steps in order unless the user specifies otherwise
3. For each technique, consider applicability to the current target/context
4. Track which checklist items have been completed
5. Suggest next steps based on findings

Full Methodology

OAuth Security Testing

Shortcut

• Check for improper redirect validation (open redirects)
• Test state parameter manipulation/absence
• Manipulate OAuth flows to bypass authentication
• Try URL path traversal in redirect_uri
• Hunt for client secret leakage in source code/repos
• Look for improper scope validation

Mechanisms

• OAuth 2.0 authorizes limited access to resources via tokens; pair with OIDC for identity.
• Core Flows:
• Authorization Code (with PKCE for public clients)
• Client Credentials (service-to-service)
• Avoid Implicit and ROPC where possible
• Key Components:
• Resource Owner (user)
• Client (third-party app)
• Authorization Server (issues tokens)
• Resource Server (hosts protected resources)
• Tokens (access and refresh)
• Hardening Extensions:
• PAR (Pushed Authorization Requests), JAR (Request Objects), JARM (JWT-secured responses)
• Sender‑constrained tokens (DPoP, mTLS)
• private_key_jwt or mTLS client authentication for confidential clients

OAuth/OIDC Considerations

• PKCE everywhere: Even with confidential clients/native apps; code_verifier must be required and validated.
• Nonce/state binding: For OIDC, ensure nonce is present and matched; state should be unguessable and tied to session.
• redirect_uri exact match: Enforce exact string match against pre-registered allowlist; no wildcards/path traversal.
• aud/azp/iss enforcement: Validate tokens strictly, including clock skew and JWKS kid rotation behavior.
• Front-channel logout/login CSRF: Validate logout CSRF; defend forced login to attacker accounts.
• ID Token vs Access Token: APIs must not accept ID tokens; check token_type and audience.
• Device Code & CIBA: Validate polling rate limits, code expiry, and binding of device/user codes.
• Refresh Token Rotation: Enforce reuse detection and global invalidation chains.
• PAR/JAR/JARM: Use to pin exact redirect_uri and inputs and to protect front-channel parameters.

OAuth 2.1 Updates

• Implicit Flow Deprecated: Authorization servers should not support response_type=token
• Password Grant Deprecated: ROPC (Resource Owner Password Credentials) considered insecure
• PKCE Mandatory: Required for all OAuth clients including confidential clients
• Exact Redirect URI Matching: No more substring or prefix matching allowed
• Refresh Token Sender Constraint: Refresh tokens should be sender-constrained via DPoP or mTLS

Financial-grade API (FAPI) Security

FAPI 1.0 Advanced Profile

• Signed Request Objects (JAR): Authorization requests as signed JWTs
• Hybrid Flow: Uses response_type=code id_token for additional security
• MTLS Client Authentication: Certificate-bound tokens
• JARM: JWT-secured authorization response mode
• Request Object Encryption: Sensitive parameters encrypted

FAPI 2.0 Security Profile

• Pushed Authorization Requests (PAR): POST request parameters to dedicated endpoint
• DPoP (Demonstrating Proof-of-Possession): Token bound to client's key pair
• Client Authentication: private_key_jwt or MTLS required
• Grant Management: Rich authorization requests and grant management API

graph TD
    User[Resource Owner] -->|Initiates flow| Client
    Client -->|Authorization Request| AuthServer[Authorization Server]
    AuthServer -->|Authentication| User
    User -->|Approves access| AuthServer
    AuthServer -->|Authorization Code| Client
    Client -->|Code + Client Secret| AuthServer
    AuthServer -->|Access Token| Client
    Client -->|Access Token| ResourceServer[Resource Server]
    ResourceServer -->|Protected Resource| Client

    style User fill:#b7b,stroke:#333,color:#333
    style Client fill:#aae,stroke:#333,color:#333
    style AuthServer fill:#9f9,stroke:#333,color:#333
    style ResourceServer fill:#e9a,stroke:#333,color:#333

Hunt

• Intercept OAuth flows with proxy (Burp/ZAP)
• Manipulate redirect_uri parameters
• Remove/tamper state parameter
• Test PKCE implementations
• Inspect token handling in browsers
• Check for client secret leakage
• Analyze scope handling logic
• Test account linking/unlinking
• Review token validation procedures
• Examine refresh token security

Native/Mobile

• Verify App Links/Universal Links to prevent hijacking callbacks.
• Ensure OAuth proxy components in mobile apps validate issuer and JWKS; do not ship client secrets in binaries.

SPA/Browser

• Use Authorization Code + PKCE; avoid Implicit/Hybrid unless justified.
• Store tokens in memory; if cookies are used, set __Host- prefix with HttpOnly; Secure; SameSite.

Authorization Code Flow

• Initial authorization request has response_type=code
• Request format: /authorization?client_id=12345&redirect_uri=https://client-app.com/callback&response_type=code&scope=openid%20profile&state=ae13d489bd00e3c24
• Callback contains authorization code: /callback?code=a1b2c3d4e5f6g7h8&state=ae13d489bd00e3c24
• More secure, backend exchanges code for tokens

sequenceDiagram
    participant User
    participant Client
    participant AuthServer as Authorization Server
    participant API as Resource Server

    User->>Client: 1. Click "Login with Service"
    Client->>AuthServer: 2. Authorization Request (response_type=code)
    AuthServer->>User: 3. Login & Consent
    User->>AuthServer: 4. Approves Access
    AuthServer->>Client: 5. Redirect with Authorization Code
    Client->>AuthServer: 6. Token Request (code + client_secret)
    AuthServer->>Client: 7. Access & Refresh Tokens
    Client->>API: 8. API Request + Access Token
    API->>Client: 9. Protected Resource

Implicit Flow

• Initial authorization request has response_type=token
• Request format: /authorization?client_id=12345&redirect_uri=https://client-app.com/callback&response_type=token&scope=openid%20profile&state=ae13d489bd00e3c24
• Access token returned directly in URL fragment: /callback#access_token=z0y9x8w7v6u5&token_type=Bearer&expires_in=5000&scope=openid%20profile&state=ae13d489bd00e3c24
• Higher vulnerability potential due to frontend token handling

sequenceDiagram
    participant User
    participant Client as Client (Browser)
    participant AuthServer as Authorization Server
    participant API as Resource Server

    User->>Client: 1. Click "Login with Service"
    Client->>AuthServer: 2. Authorization Request (response_type=token)
    AuthServer->>User: 3. Login & Consent
    User->>AuthServer: 4. Approves Access
    AuthServer->>Client: 5. Redirect with Access Token in Fragment
    Note over Client: Token stored in browser
    Client->>API: 6. API Request + Access Token
    API->>Client: 7. Protected Resource

Vulnerabilities

• Improper redirect_uri validation
• Open redirects