Identity authenticity
Determine whether tokens, assertions and authorization responses are bound to the correct issuer, audience, client, flow and session.
Federated identity testing
Federation failures rarely sit in one endpoint. They emerge from the relationship between identity provider, client, redirect handling, token validation, account linking, tenant discovery and session state. We test SAML, OAuth 2.0 and OpenID Connect implementations as complete trust flows, with controlled identities and evidence that developers can reproduce safely.
Decision firstScope the question and relying audience
Safe authorizationSet access, exclusions and stop conditions
Validated evidenceSeparate facts, impact and uncertainty
Owned next moveRemediate, retest or accept explicitly
The buyer decision
A standards-compliant library does not make the integrated product secure by itself. The application still decides which issuer to trust, how to bind a response to a browser session, whether an email identifies the same person, which tenant or role to assign and how long the resulting session remains useful. We challenge those product decisions as well as protocol mechanics.
The engagement fits teams launching enterprise SSO, supporting multiple identity providers, adding social login, brokering identity across tenants or exposing APIs to third-party clients. Results distinguish protocol vulnerabilities, dangerous integration choices and hardening opportunities so engineering can address the correct layer.
Determine whether tokens, assertions and authorization responses are bound to the correct issuer, audience, client, flow and session.
Test whether linking, email matching, just-in-time provisioning or invitation flows can attach an attacker to another identity.
Verify scopes, claims, tenant mapping, roles and consent do not grant more access than the user or client should hold.
Assess logout, revocation, reauthentication and session transitions after identity, role or membership changes.
Assessment coverage
We scope only the flows and parties you authorize. The exact test matrix depends on whether the application is a client, resource server, identity provider or broker.
SAML service provider and identity provider integrations
OAuth 2.0 authorization and token flows
OpenID Connect authentication and discovery
Redirect URI, state, nonce and PKCE handling
Issuer, audience, signature and key validation
Account linking, JIT provisioning and domain discovery
Claims, scopes, roles and tenant assignment
Logout, token revocation, refresh and session lifecycle
Risk paths
Test cases are adapted to the enabled grants, clients and providers. We avoid generic payload dumping and concentrate on reachable trust failures.
Test whether state, nonce and browser-session binding prevent a victim from being logged into or linked with an attacker-controlled identity.
Assess PKCE enforcement, redirect handling and code binding for public and confidential clients.
Challenge multi-provider and multi-tenant flows for mix-up, discovery and audience validation errors.
Attempt to reuse access tokens, ID tokens, assertions or tokens from another client, environment or resource.
Probe exact redirect matching, nested redirects, parser differences and post-login return destinations.
Test email matching, invitation acceptance, verified-domain assumptions and link/unlink workflows.
Validate that group, role, tenant and administrative claims are issued and consumed with the expected authority.
Evaluate whether logout, disabled users, removed memberships and rotated secrets invalidate useful access as intended.
Scope architecture
A flow inventory prevents a successful test of one client from being mistaken for assurance over the entire identity estate.
| Flow component | Representative questions | Required context |
|---|---|---|
| Authorization request | Are redirect URI, response type, state, nonce and PKCE requirements enforced for this client? | Client type, enabled flows and registered callbacks |
| Response validation | Are issuer, audience, signature, time, nonce and code bindings checked against trusted configuration? | Provider metadata, libraries and expected claims |
| Account resolution | How are subject, email, domain, invitation and existing account records reconciled? | Linking rules, JIT behavior and controlled identities |
| Authorization | How do scopes, groups, claims and tenant context become application permissions? | Role model, API resources and consent design |
| Lifecycle | What happens when access is revoked, a user changes tenant, a key rotates or an administrator disables SSO? | Session model, refresh behavior and recovery flows |
Delivery process
Every phase preserves scope, authorization, evidence quality and a practical path to closure.
Document providers, clients, resource servers, redirect endpoints, signing keys, claims, account states and administrative control points.
Prepare authorized accounts across relevant providers, tenants and roles. Define token handling, evidence minimization and escalation rules.
Test browser, back-channel and API behavior using protocol-aware changes. Confirm exploitability and impact without retaining unnecessary tokens.
Map each finding to application logic, protocol configuration, provider policy or library usage and retest the corrected trust decision.
Deliverables
Deliverables are tailored to the agreed audience. Technical evidence remains access-controlled; summaries keep the tested scope, date and limitations attached.
Federation and authorization trust-flow map
Client, provider and redirect inventory
Validated account-takeover and authorization findings
Sanitized request, response and token evidence
Account-linking and tenant-mapping observations
Protocol-specific remediation guidance
Untested flows and reliance assumptions
Retest status for agreed corrections
Service fit
Evidence boundaries
RFC 9700 documents current OAuth 2.0 security best practice, including defenses such as exact redirect matching and PKCE. OpenID Connect Core defines security considerations for authentication built on OAuth. We use applicable requirements to inform test cases, then validate how the deployed parties actually behave. Referencing a standard does not make this a conformance certification.
The report covers only authorized providers, clients, flows, environments and account states. Identity ecosystems change when providers rotate keys, administrators alter claims or new clients are registered. Owners receive regression cases and should repeat critical checks after material federation changes.
Common questions
Yes, when those protocols and flows are included in the authorized scope. They solve different problems and require different test cases. We inventory each provider, client and relying party so a result for one integration is not incorrectly generalized to another.
The standard approach uses controlled identities and avoids configuration changes. We agree rate limits, prohibited actions and escalation contacts in advance. Any test with a realistic chance of affecting production authentication requires explicit approval and a rollback-ready operating plan.
Often we can test external behavior without receiving sensitive keys. If a deeper review requires secrets or configuration exports, we minimize access and use an agreed secure transfer method. Do not send tokens, private keys or credentials in the initial email or scope form.
Yes. Those product decisions are common sources of material identity risk and are included when relevant. We test how subject identifiers, emails, verified domains, invitations and claims become application accounts, tenants and roles.
No. The engagement is a security assessment of specified implementations and flows. Applicable standards inform coverage, but the report does not certify protocol conformance or guarantee the security of untested clients, providers, future changes or third-party services.
Related services
Next step
Share the decision, systems, important roles or environments, access constraints and deadline. Begin without credentials or protected evidence; NDA and secure transfer can be agreed before sensitive material is exchanged.