AZVulnerability family
Object, role, account, and tenant access-control design. 34 disclosures · 83 related references · 10 diagrams.
Connected collection
Mamba received a distinct USD 37,604.40 award. The case illustrates why temporary processing output needs explicit access isolation, with retrieval dependencies kept separate from standalone impact.
Firefly received a separately identified USD 60,000 award. The case illustrates how delegated worker authority and storage permissions can diverge from caller authorization.
A researcher reports USD 13,337 for a device-grant authorization-boundary failure.
A researcher organization reports a $150,000 base award for an exposed service identity with excessive downstream integration authority.
A researcher reports a $75,000 base award for a distinct Application Integration report.
Ordinary authenticated customers could read internal support activity; the researcher reports a USD 14,337 award.
A researcher reports a $15,000 Google bounty for a Gemini Enterprise integration issue affecting persistent assistant memory.
A Google-rewarded report connected an adjacent CI misconfiguration with insufficient separation of automation trust, creating a potential Angular supply-chain impact.
A Kestrel request-framing report earned a researcher-reported USD 10,000 award.
GitHub awarded USD 10,000 for a GHES comparison feature that exposed limited code across repository permission boundaries (vendor-report).
A support-system misconfiguration allowed external access to internal Confluence documentation, including limited content modification. One report received USD 12,500 split between two researchers.
A researcher-reported USD 30,000 award illustrates a gap between approval of a contribution and selection of the code executed.
One Actifio report earned USD 10,000 after a collaboration multiplier.
A framework upgrade exposed private contributor attributes in public report responses. The vendor confirmed a USD 25,000 award; the public report supplies the underlying serialization and test-normalization explanation.
A monetized-channel account could access another monetized creator’s private email through inconsistent cross-API authorization. The researcher reports a USD 20,000 award.
One recovery report received USD 35,000 across two award events.
One cross-product privacy report received USD 3,133 and a USD 7,500 adjustment, totaling USD 10,633.
HackerOne awarded USD 25,000 in 2022 for private-program GraphQL metadata exposure; the report became public in January 2025 (primary).
Patchstack confirmed a USD 14,400 Zero Day award for an authentication-boundary flaw.
An isolation misconfiguration exposed runner-image build infrastructure; the researcher reports a $20,000 payment.
The researcher documents USD 14,500 for one report: a 10,000 base bounty and two event bonuses.
Meta confirms a USD 44,250 total award, including bonuses, for this Quest OAuth account-access report.
Meta confirmed a $30,000 base award for an Instagram application-token exposure; additional researcher-listed bonuses are excluded from the recorded amount.
A reference-validation flaw crossed the GitHub Actions trust boundary and earned USD 25,000.
The second fork-collaboration report received its own USD 10,000 award.
The first fork-collaboration finding in this article earned USD 20,000.
Bhavuk Jain reports being paid $100,000 for this finding. The recorded USD denomination is a contextual inference from Apple's May 2019 program documentation, published about one year before the May 2020 disclosure.
A USD 25,000 researcher-reported award illustrates how differing framework and controller assumptions can remove an OAuth consent boundary.
Shopify confirms a $20,000 award for unauthorized collaborator access caused by automatic account conversion.
HackerOne awarded $12,500 for internal attachments exposed through report export in 2016. Its enduring lesson for 2026 applications is that export and interactive views must enforce the same visibility policy.
A 2013 researcher disclosure reports a $20,000 award for unauthorized recovery-phone binding. Its enduring lesson for 2026 applications is that recovery-factor possession and account-change authority require separate…
The USD 32,500 researcher-reported award illustrates the distinction between message origin and disclosure authority.
An authenticated media-editing operation exposed another user’s prompts and generated content because object ownership was not enforced. The researcher reports a USD 10,000 award.
A researcher-reported USD 30,000 award documents a cross-product account-linking trust failure.
Connected collection
Explains the deployment assumption behind centralized GraphQL federation controls: internal subgraphs accept traffic only through the router. Federation coordination remains available even when ordinary client…
Astro's base-path removal accepted a textual prefix without establishing a complete path segment. Routing and authorization middleware could consequently disagree about the requested resource. The maintainer bounds the…
CVE-2026-49443 concerns writable identity-mapping fields in API serializers. Delegated connection-management authority could alter which local user or group a source identity represented. The maintainer reports…
CVE-2026-44681 describes an OIDC error path selecting a response destination before client and destination validation. The maintainer reports an unauthorized browser redirect, explicitly excluding direct disclosure of…
Use this guide to review machine identity design: favor short-lived role credentials for workloads, limit permissions to required actions and resources, and retire unnecessary access. It also explains policy validation,…
The maintainer disclosure for CVE-2026-53518 identifies separated reading and deletion of a single-use OAuth authorization record. Under concurrent processing, multiple successful consumers could receive independent…
CVE-2026-53516 concerns implicit linking to an unverified local account. Provider-side email verification was allowed to confer legitimacy on previously stored local credentials. The resulting merged identity could…
Personal SCIM providers could lack an owner, while management checks rejected mismatched ownership only when an owner existed. Missing identity binding therefore admitted unrelated authenticated users. The maintainer…
GHSA-4f45-qmjf-82cv describes account-access links whose configured seconds were interpreted as days. The maintainer reports unused email-verification, password-reset and new-user setup tokens surviving their intended…
CVE-2026-55429 concerns a provisioning update-or-insert operation that could change an existing workspace application’s ownership relationship without checking the existing workspace. The maintainer describes potential…
Zafran’s tracing case, CVE-2026-41947, explains that console authentication did not bind configuration changes to the application’s tenant. Traces contain prompts and responses, so changing telemetry routing also…
GHSA-p623-wgx3-wxp8 describes service overrides committing cleanup before their superclass checked permission. Rejected mutations could therefore disconnect automation links, remove attribution metadata or invalidate…
Research on CVE-2026-71505 describes inconsistent authorization between reading and modifying company portal accounts. A role-level check did not establish authority over the particular company. In researcher-owned…
GHSA-v9w4-gm2x-6rvf describes public file-sharing authority outliving its owner's permissions. Creating a share required sharing and download rights, but subsequent public access did not revalidate those rights. The…
CVE-2026-39351 concerns related-document expansion in a REST response. Frappe loaded linked records without checking the caller’s permission and serialized them without field filtering. The research identifies…
Wiz's CVE-2026-3854 research examines a data-to-authority boundary in GitHub's backend. User-controlled operation metadata reached downstream services as trusted configuration. Code execution was demonstrated on…
Google's Approved pagination guidance separates a continuation position from permission to read the collection. Page tokens must be opaque and convey no authorization; each request still requires authorization.…
Google's Approved field-mask guidance defines which resource fields participate in an update. Services must ignore output-only input whether selected directly or through a containing field. Supporting field-behavior…
A GraphQL-Ruby execution-engine path converted a resolver authorization exception into permission to continue. Research demonstrates a denied resolver running and returning a fixture value, while the legacy engine…
GHSA-7qfj-82q8-frw6 describes inconsistent account-disable enforcement across authentication methods. Existing browser or remembered sessions could retain API authority because refreshed permissions did not also…
GHSA-v8jx-vq6p-jq52 describes an API permission gap between viewing a submission and changing contact authorship. The maintainer reports that reviewers and program-committee members could obtain author-level access to…
The maintainer confirms that project-level MCP authentication did not authorize later file reads. Resource handlers reached storage without retaining user and project restrictions. A local two-user demonstration showed…
The advisory contrasts denied access to a private agent with accepted changes to its attached files. Upload handling omitted the agent permissions enforced elsewhere, allowing unauthorized context and search-file…
CVE-2026-31944 describes an MCP OAuth callback that trusted cached initiator identity without authenticating the returning browser or checking identity continuity. The advisory describes third-party credentials being…
The MCP registry prepared decrypted configuration for internal use, and response handlers returned that representation to viewers without removing secrets. Object visibility consequently became credential disclosure…
The maintainer advisory attributes cross-user knowledge-base file removal to an omitted ownership restriction in a database mutation. Its reported demonstration shows one deletion result. The boundary is between being…
Form elicitation excludes secrets. URL elicitation places sensitive interactions outside the MCP client and model context, with the requesting server and destination visible to the user. Agreeing to open the interaction…
Server permission challenges shape what a general-purpose MCP client requests. Broad discovery metadata and accumulated scopes can enlarge delegated authority. Token scopes still require application-side authorization.
Microsoft Graph documentation distinguishes batch-envelope success from individual outcomes. Its example includes permission denials inside a successful batch response. Member results can arrive in a different order and…
Tachyon’s 2026 account of CVE-2025-14297 describes resource permissions depending on incomplete interface registration. Authentication could succeed while a missing authorization mapping permitted resource access. The…
GHSA-qmh4-5v7g-42jq concerns pending payment state being accepted as authority to grant account credit before provider-confirmed settlement. The account update was atomic, but that concurrency property did not establish…
CVE-2026-54305 concerns authenticated Dynamic Credentials operations missing workflow and credential ownership or scope checks. Maintainers report unauthorized credential metadata access, OAuth identity replacement and…
CVE-2026-33665 describes local-account linkage that trusted a matching LDAP email attribute. The maintainer reports persistent access to the linked account, including administrator authority, even after the directory…
The maintainer reports that initial OAuth authorization preserved resource-specific consent, while refresh grants checked registration without preserving that binding. A client could consequently receive authority over…
An evaluation of 27 services and a 100-participant user study examines missing context, explicit consent and post-login control in cross-device authentication. Conceptual boundary: an already trusted device may approve…
Explains how server rendering changes data-access assumptions. A dedicated server-side data layer can centralize authorization and expose only fields required by the interface. Server Actions need their own caller and…
CVE-2026-41574 describes provider adapters converting email presence or fallback profile attributes into a verification claim. An account-linking consumer then treated that normalized claim as ownership evidence. The…
Defines authorization in terms of subject, object, operation and environmental attributes evaluated against policy. Enterprise considerations connect business rules to machine-enforced decisions, attribute authorities…
CVE-2026-71316 documents an authorization mismatch between server-rendered HTML and extracted data. A shared path-keyed payload cache omitted requester identity and returned data before current-request guards. The…
GHSA-wr5q-7wxw-x568 describes client-address metadata acquiring authority to waive authentication without verified transport-peer and intermediary provenance. The maintainer reports unauthenticated access within the…
GHSA-xwmw-prc4-v3cr describes MCP-delegated tokens accepted by broader application APIs because issuer validation did not establish audience authority. Client authorization also lacked explicit consent. The maintainer…
CVE-2026-87015 concerns late-bound connection state: tool callables retained their own headers but read a shared cookie variable after connection processing finished. A maintainer-described controlled observation…
CVE-2026-59219 documents inconsistent session invalidation: HTTP authentication consulted revocation state, while realtime authentication checked only token signature and expiry. The reported local comparison shows…
GHSA-2rr4-q6pg-m5g3 documents a shared-policy input mismatch: browser sign-in supplied identity-token claims, while token exchange supplied only provider user information. Missing role evidence preserved an old role.…
CVE-2026-61709 describes incorrect authorization-policy evaluation in user enumeration. Under a particular composition of wildcard membership, exclusion and intersection, a user denied by one policy component could…
Authorization correctness includes the age of relationship state used for a decision. OpenFGA documents a latency-oriented mode that can reuse cached results and a higher-consistency mode that bypasses the cache. With…
Defines a high-security OAuth profile with coordinated requirements for confidential clients, authorization servers, and resource servers. Connects sender-constrained tokens and authorization-request integrity with the…
GHSA-33jq-x32c-3ccw describes webhook authority surviving deletion of its creator. Account cleanup omitted webhook subscriptions, while delivery trusted their enabled state. The maintainer-published report describes a…
Explains account-bound recovery challenges, limited lifetime and reuse, consistent responses, attempt controls, notifications and post-reset session handling.
Practical design guidance covering least privilege, deny-by-default behavior, consistent per-request decisions, failure handling, logging, and authorization regression tests.
OWASP explains origin checks, expected message formats and treating exchanged content as data. These controls address different assumptions at browser communication boundaries.
Explains why model-generated content remains untrusted when passed to browsers, databases or backend functions. The relevant boundary is the consuming component: plausible model text must not acquire executable meaning…
Defense-in-depth guidance for LLM applications that consume untrusted content or invoke tools. Covers data provenance, least privilege, action authorization, monitoring, and the limitations of guardrails.
Examines security assumptions around retrieval-augmented generation, including access to embeddings, cross-context disclosure and integrity of imported knowledge. Shared retrieval infrastructure must preserve the…
Distinguishes application-issued session identifiers from client-selected values. Explains renewing identifiers at login and other privilege changes, retiring previous identifiers, and separating anonymous tracking from…
Explains operation-specific approval: show significant transaction details, preserve authorized data, enforce valid state transitions and recheck authorization at execution.
Research on CVE-2026-29000 describes an authentication boundary failure: successful decryption could allow claims to become an authenticated profile without mandatory signature validation. The researcher reports…
CVE-2026-59965 describes authenticated plugin endpoints invoking a privileged server interface without preserving collection-level authorization. Payload's Local API skips access checks by default, so a session check…
The maintainer traces duplicate downgrade credits to an eligibility check separated from the later balance change, without transactional isolation. A pending-operation guard did not protect the whole transition. The…
The API draft distinguishes permission for the top-level and embedded site pair from activated access in a document and cookie eligibility on each request. Secure context, origin, embedding-policy and sandbox conditions…
CVE-2026-59151 concerns token issuance selecting a tenant from an asserted email domain instead of retaining the validated SAML configuration. Maintainers describe potential cross-tenant account takeover. Their…
GHSA-8r6w-3qq5-4p4r describes an authorization mismatch between the Panel and Wings. Tokens established an authenticated user and server context without adequately separating operation purpose. The maintainer-published…
Compares browser-only OAuth clients, token-mediating backends, and backend-for-frontend architectures through their different token-custody and session boundaries. Separates protection of token material from the…
Consensus guidance updating OAuth's security model with deployment experience, stronger protocol requirements, and deprecated insecure patterns. A primary reference for identity integration reviews.
The maintainer describes inconsistent escaping between XML attribute and element-text contexts during SAML assertion generation. User-controlled profile values could change assertion structure before the identity…
CVE-2026-26004 concerned a missing organization constraint in event retrieval. A permission check on the active organization did not establish ownership of the requested object. GitHub Security Lab reports…
Learn to assess software supply-chain assurance using distinct source and build tracks. The build track progresses from recording provenance to authenticated hosted builds and stronger platform isolation. Build…
CVE-2026-94462 concerns a guest-cart ownership transition that required a signed-in customer but omitted the cart-possession check enforced by sibling operations. Account authentication and object lookup were treated as…
CVE-2026-25757 concerns completed guest orders. The access decision treated absence of an account owner as sufficient permission, while lookup did not require the separate order token. GHSL identifies the flaw in tested…
CVE-2026-75523 describes diagnostic URI masking that removed inline credentials but preserved query contents. The failed assumption was that sanitizing one URI component made the whole representation safe for secondary…
CVE-2026-31820 concerns client-supplied component action arguments used to load objects without ownership checks. Property checksums did not cover those arguments. GHSL describes cross-customer address disclosure in…
The advisory distinguishes ownership of an order from authority over its payment operations. Customer-context requests were constrained by ownership but not by operation. A connected payment provider could therefore…
CVE-2026-31834 separates authority over an account from authority to grant its roles. Research on Umbraco CMS 17.2.0 found that group-membership changes checked access to target users but omitted restrictions applied by…
Studies account linking across multi-app integration platforms and identifies inconsistent app identity as a trust-boundary problem. Proposes app-specific authorization-context binding.
Studies a cross-layer authentication failure: shared TLS session-ticket infrastructure can preserve cryptographic session state without preserving the intended virtual-host identity and client-authentication policy. The…
A qualitative lab study of 31 participants across three services found that unclear passkey labels, notifications and recovery interfaces obstructed complete account remediation. Conceptual boundary: proof that a…
Order-level scoping did not carry into globally loaded payment, refund and fulfillment objects. A channel-limited administrator could affect another channel through child-object operations. The advisory contrasts scoped…
GHSA-7wjf-49rm-77gx describes incoming session identifiers being used as storage keys without validation. Constrained generation did not constrain values returning from the client. The maintainer reports read, overwrite…
GHSA-jp29-jrxc-92vf describes a task-search branch that trusted a saved favorite without checking current project access. Favorite records survived share revocation, allowing previously authorized collaborators to keep…
Defines browser-provided request context covering site relationship, destination, mode, and user activation. Explains how redirect history affects that context and why context-dependent responses need matching cache…
Defines browser-feature availability through inherited restrictions, document declarations and frame delegation. Feature defaults govern undeclared cases; a child cannot restore authority disabled by its parent.
Defines how opener policies affect browsing-context separation during navigation. The standard expressly distinguishes severing an opener relationship from a robust boundary between same-origin documents: storage,…
GHSL-2026-049 describes a ticket asset serializer that overrode a permission-aware base implementation without preserving its group-access check. GitHub Security Lab reports confidential ticket and associated-user…
Connected collection
Editorial conceptual model derived from the linked cases and official guidance; not a vendor architecture diagram or an exploitation sequence.
Editorial conceptual model derived from the linked cases and official guidance; not a vendor architecture diagram or an exploitation sequence.
Editorial conceptual model derived from the linked cases and official guidance; not a vendor architecture diagram or an exploitation sequence.
Original defensive model combining OWASP messaging and authorization guidance with the linked historical cases. These are independent design checks, not a vendor patch diagram or an operational reproduction.
Editorial conceptual model combining the Angular case with SLSA build guidance. The case supports separating automation trust and shared build state; the consumer verification gate is a general design synthesis, not a…
Original conceptual model for a view that requires multiple independently protected sources. Permission to use one source cannot authorize another source. The convergence is an AND requirement: both source decisions…
Original editorial defensive synthesis of the Instagram embedding disclosure and OWASP authorization guidance. The researcher attributes the disclosure to error handling that retrieved protected content under an…
Editorial conceptual model derived from the linked cases and official guidance; not a vendor architecture diagram or an exploitation sequence.
Editorial conceptual model: assumes an application with server-side privileged data and a browser consumer. Next.js guidance supports server authorization and minimal client-visible contracts; OWASP LLM05 supports…
Editorial conceptual model derived from the linked cases and official guidance; not a vendor architecture diagram or an exploitation sequence.
Related learning follows the topic crosswalk or an explicit diagram relationship. It does not classify a resource as a finding. Topics overlap, so their counts should not be added together.
GitHub snapshot 2026-10-04
53796974ace8 · JSON exports & schemas · CC BY 4.0 content / MIT software
| / | Focus search on this page |
| Ctrl K | Search everything (command palette) |
| 1-9 | Library, Generator, Playbooks, Gadgets, Checklists, Payloads, Bypasses, Utilities, Reports |
| 0 | AI / MCP connector |
| j / k | Move selection down / up |
| Enter | Expand / open selected |
| c | Copy primary command of selected |
| f | Toggle favorite on selected tool |
| Esc | Clear search / close |