vulns.co
/
GKData.io MCP and API Buy me a coffee

Vulnerability family

Authorization and tenant boundaries

Object, role, account, and tenant access-control design. 34 disclosures · 92 related references · 11 diagrams.

Disclosures

Related learning

Apollo GraphQL Architecture Guide

Apollo Federation: preserving the router-to-subgraph boundary

Apollo's guidance says internal GraphQL subgraphs must accept traffic only through the router for centralized security controls to hold. Disabling ordinary client introspection does not remove the coordination features…

Reviewed 2026-10-03

Astro Maintainer Advisory

Astro: routing and authorization must agree on resource identity

Astro's routing and permission middleware could interpret the same path differently because base-path removal did not check for a complete path segment. The maintainer limits the resulting authorization bypass to…

Reviewed 2026-10-03

authentik Maintainer Advisory

authentik: source-mapping edits carry identity-rebinding authority

Permission to manage an authentik source connection could also let a user change which local account or group it represented. The CVE-2026-49443 advisory attributes this to identity-mapping fields that the API allowed…

Reviewed 2026-10-03

Amazon Web Services Implementation Guide

AWS IAM security best practices for workload identities

AWS recommends short-lived role credentials for workloads, permissions limited to the actions and resources they need, and removal of unnecessary access. Its guide also covers policy validation, access reviews and…

Reviewed 2026-10-02

Axios Maintainer Advisory

Axios: URL construction does not establish destination authority

An Axios base URL does not by itself restrict where requests can go: an absolute URL can select a different destination. The CVE-2025-27152 advisory describes possible server-side request forgery when applications…

Reviewed 2026-10-05

Better Auth Maintainer Advisory

Better Auth SCIM: absent ownership must not grant shared authority

Better Auth's maintainer reports that personal SCIM providers without an owner could be managed by unrelated signed-in users. The permission check rejected a mismatched owner only when an owner was recorded. Reported…

Reviewed 2026-10-03

Bugsink Maintainer Advisory

Bugsink: time-unit consistency in account-access token expiry

Bugsink interpreted token lifetimes expressed in seconds as days. Its maintainer reports that unused email-verification, password-reset and new-user setup links therefore stayed valid beyond their intended expiry.

Reviewed 2026-10-03

Directus Maintainer Advisory

Directus: denied mutations must leave dependent state unchanged

Directus could reject a change after cleanup had already modified application state. Its maintainer advisory describes disconnected automation links, removed attribution metadata and invalidated permission caches…

Reviewed 2026-10-03

File Browser Maintainer Advisory

File Browser: existing shares must follow current owner permissions

File Browser's maintainer reports that existing public share links still allowed downloads after the owner's permissions were revoked in version 2.62.2. Creating the share required both sharing and download permissions,…

Reviewed 2026-10-03

Verbb / Formie Maintainer Advisory

Formie: integration settings need operation and attribute authority

Formie's maintainer reports that weak permission checks and overly broad settings changes could expose stored integration credentials. Callers could change the destination used with those credentials, and the advisory…

Reviewed 2026-10-04

GitHub Security Lab Research Paper

Frappe: linked data must preserve document and field permissions

Frappe could include documents a caller was not allowed to read when expanding linked records in an API response. GitHub Security Lab's CVE-2026-39351 research found that these records were loaded without permission…

Reviewed 2026-10-03

Google Implementation Guide

Google API field masks: preserve server-owned and immutable state

Google's approved field-mask guidance defines which fields an update selects, while separate field rules determine what may change. Services must ignore output-only input even when a containing field is selected; the…

Reviewed 2026-10-04

GitHub Security Lab Research Paper

GraphQL-Ruby: authorization exceptions must stop execution

GraphQL-Ruby could continue running a resolver after its authorization check raised an exception. GitHub Security Lab demonstrated the denied resolver returning a test value in the newer execution engine, while the…

Reviewed 2026-10-03

HotCRP Maintainer Advisory

HotCRP: separate submission visibility from authorship authority

HotCRP allowed permission to view a submission to extend to changing its contact authorship. The maintainer reports that reviewers and program-committee members could gain author-level access to submissions they could…

Reviewed 2026-10-03

LibreChat Maintainer Advisory

LibreChat: agent edit authority must cover attached context

LibreChat could deny access to a private agent yet accept unauthorized additions to its context and search files. The maintainer advisory attributes this to upload handling that omitted the agent permission checks used…

Reviewed 2026-10-03

Model Context Protocol Technical Standard

MCP elicitation: consent, credential custody and completion

The MCP specification keeps secrets out of form elicitation and places sensitive URL interactions outside the client and model context, with the requesting server and destination visible to the user. Agreeing to open an…

Reviewed 2026-10-04

Model Context Protocol Architecture Guide

MCP scope selection: progressive consent and accumulated authority

MCP's official guidance explains how a server's permission requests influence the permissions a general-purpose client asks the user to grant. Broad discovery metadata and permissions accumulated over time can expand…

Reviewed 2026-10-03

Tachyon Research Paper

MLflow: authorization must survive alternate resource interfaces

MLflow could authenticate a user yet allow restricted resource access when an interface lacked an authorization mapping. Tachyon's 2026 account of CVE-2025-14297 demonstrates restricted artifact reads and describes…

Reviewed 2026-10-03

MythicalDash Maintainer Advisory

MythicalDash: payment evidence must establish credit entitlement

MythicalDash could grant spendable account credit from a pending payment before the provider confirmed settlement. Its maintainer advisory reports unpaid credits in a controlled deployment. The balance update was…

Reviewed 2026-10-04

n8n Maintainer Advisory

n8n Dynamic Credentials: authorize credential lifecycle operations

n8n's maintainer advisory reports missing ownership and scope checks in authenticated Dynamic Credentials operations (CVE-2026-54305). The reported effects include unauthorized access to credential metadata, replacement…

Reviewed 2026-10-03

n8n Maintainer Advisory

n8n token exchange: preserve the issuer namespace

n8n's maintainers report that token exchange matched external identities by subject without keeping track of the issuer (CVE-2026-59208). When the feature was enabled with multiple trusted issuers, this could associate…

Reviewed 2026-10-05

n8n Maintainer Advisory

n8n: directory-attribute authority in durable account linking

n8n's maintainer advisory describes account linking that treated a matching LDAP email attribute as proof of account ownership (CVE-2026-33665). It reports that access to the linked account, including administrator…

Reviewed 2026-10-03

Internet Society / NDSS Symposium Research Paper

Cross-device authentication: bind informed consent to session authority

Approving a login from a trusted device does not, by itself, show that the user understands and wants the new session. This research evaluates 27 services and includes a 100-participant user study of missing context,…

Reviewed 2026-10-03

Next.js / Vercel Implementation Guide

Next.js data security: server authorization and client-visible data

Next.js guidance recommends a server-side data layer that keeps authorization checks together and returns only the fields the interface needs. Server Actions still need checks on the caller and requested resource;…

Reviewed 2026-10-03

Nhost Maintainer Advisory

Nhost: provider adapters must preserve identity-claim evidence

Nhost's maintainer advisory reports that provider adapters treated an email address or fallback profile attribute as a verified identity claim (CVE-2026-41574). Account linking then accepted that claim as proof of…

Reviewed 2026-10-03

National Institute of Standards and Technology Architecture Guide

NIST SP 800-162: attribute authority and policy traceability

NIST explains how access policies use attributes of the user, resource, requested action and environment to decide whether to allow access. It connects business rules to software-enforced decisions and explains why…

Reviewed 2026-10-03

Nuxt Maintainer Advisory

Nuxt: rendered-data caches must preserve request authorization

Nuxt's maintainer reports that a shared data cache could disclose another user's data, including to unauthenticated visitors, even while the HTML page remained protected (CVE-2026-71316). The cache used the path without…

Reviewed 2026-10-03

Obot Maintainer Advisory

Obot: preserve delegated token audience and consent boundaries

Obot's maintainer reports that tokens delegated for MCP were accepted by broader application APIs because checking the issuer did not check the intended recipient (GHSA-xwmw-prc4-v3cr). Client authorization also lacked…

Reviewed 2026-10-03

Open WebUI Maintainer Advisory

Open WebUI: credentials must bind to their destination connection

Open WebUI's maintainer reports unintended session-cookie forwarding in a controlled observation (CVE-2026-87015). Tool functions kept their own headers but read cookies from shared state after connection processing…

Reviewed 2026-10-03

Open WebUI Maintainer Advisory

Open WebUI: revocation must cross HTTP and realtime boundaries

Open WebUI's maintainer advisory describes revoked tokens being rejected over HTTP but accepted for new realtime connections in a local comparison (CVE-2026-59219). Realtime authentication checked the signature and…

Reviewed 2026-10-03

Open WebUI Maintainer Advisory

Open WebUI: preserve role-policy meaning across identity flows

Open WebUI's maintainer reports that token exchange could preserve a user's old role when role evidence was missing (GHSA-2rr4-q6pg-m5g3). Browser sign-in supplied identity-token claims, while token exchange supplied…

Reviewed 2026-10-03

OpenFGA Maintainer Advisory

OpenFGA: policy intersections must preserve explicit exclusions

OpenFGA's maintainer advisory describes excluded users incorrectly returning in user-enumeration results (CVE-2026-61709). This happened with a particular combination of wildcard membership, exclusion and intersection,…

Reviewed 2026-10-03

OpenID Foundation Technical Standard

FAPI 2.0 Security Profile

The OpenID Foundation's FAPI 2.0 profile sets coordinated OAuth security requirements for confidential clients, authorization servers and resource servers. It covers sender-constrained tokens and protection of…

Reviewed 2026-10-03

Outline Maintainer Advisory

Outline: integration authority must end with its owning account

Outline's maintainer-published report demonstrates document content still being delivered by a webhook after its administrator creator's account was deleted (GHSA-33jq-x32c-3ccw). Account cleanup left the subscription…

Reviewed 2026-10-03

OWASP Cheat Sheet Series Implementation Guide

OWASP Forgot Password

OWASP explains how to bind recovery challenges to the right account, limit their lifetime and reuse, and control repeated attempts. It also covers consistent responses, user notifications and session handling after a…

Reviewed 2026-10-07

OWASP Cheat Sheet Series Implementation Guide

Authorization Cheat Sheet

OWASP recommends granting only needed permissions, denying access by default and making consistent authorization decisions on every request. The guidance also covers failure handling, logging and regression tests for…

Reviewed 2026-10-02

OWASP Cheat Sheet Series Implementation Guide

HTML5 Security Cheat Sheet: Web Messaging

OWASP explains how to check a browser message's origin and format and handle its contents as data. These are separate controls: verifying the sender does not replace validating the message or handling its content safely.

Reviewed 2026-10-02

OWASP Foundation Training Lab

OWASP Juice Shop: Controlled Web-Security Training

OWASP Juice Shop is a deliberately insecure web application for learning how flaws arise and how to fix them. It offers challenges at different difficulty levels, optional guided tutorials and coding exercises, with…

Reviewed 2026-10-12

OWASP Gen AI Security Project Implementation Guide

OWASP LLM05:2025: generated-output consumer trust

OWASP explains why model-generated content must still be treated as untrusted input by browsers, databases and backend functions. Text should not become executable code or gain extra permissions merely because an…

Reviewed 2026-10-06

OWASP Cheat Sheet Series Architecture Guide

LLM Prompt Injection Prevention Cheat Sheet

OWASP recommends layered defenses for LLM applications that read untrusted content or use tools. The guidance covers tracking data sources, limiting permissions, authorizing actions and monitoring behavior, while…

Reviewed 2026-10-02

OWASP Gen AI Security Project Architecture Guide

OWASP LLM08:2025: retrieval permissions and knowledge provenance

OWASP explains why shared retrieval systems must keep users' permissions and data classifications intact. Its guidance covers access to embeddings, disclosure across contexts and the integrity of imported knowledge.…

Reviewed 2026-10-06

OWASP Cheat Sheet Series Implementation Guide

OWASP Session Management: privilege-transition integrity

OWASP explains why applications should issue session identifiers themselves and replace them at login or other privilege changes. Previous identifiers must be retired, and anonymous tracking kept separate from…

Reviewed 2026-10-03

OWASP Cheat Sheet Series Implementation Guide

OWASP Transaction Authorization

OWASP explains how to tie approval to a specific operation and its important details. Applications must preserve the approved data, enforce valid workflow steps and check authorization again when the operation runs.

Reviewed 2026-10-02

jhb-software / Payload plugins Maintainer Advisory

Payload: request adapters must retain caller-level authorization

A Payload plugin maintainer reports that authenticated endpoints called the privileged Local API without preserving the caller's collection permissions (CVE-2026-59965). Because the Local API skips access checks by…

Reviewed 2026-10-04

PortSwigger Training Lab

Web Security Academy: Free Online Training from PortSwigger

PortSwigger offers free web-security lessons and purpose-built interactive training environments. Topics include authentication, access control, business logic, API security and other application-security fundamentals.

Reviewed 2026-10-02

Prowler Maintainer Advisory

Prowler SAML: retain validated tenant authority

Prowler's maintainers report that token issuance chose a tenant from an asserted email domain rather than the validated SAML configuration (CVE-2026-59151). They describe possible cross-tenant account takeover. Their…

Reviewed 2026-10-03

Internet Engineering Task Force / RFC Editor Technical Standard

RFC 10017: OAuth 2.0 for Browser-Based Applications

RFC 10017 compares browser-only OAuth clients, token-mediating backends and backend-for-frontend designs by where they keep tokens and manage sessions. It distinguishes protecting token material from limiting what…

Reviewed 2026-10-03

Internet Engineering Task Force / RFC Editor Technical Standard

RFC 9700: Best Current Practice for OAuth 2.0 Security

RFC 9700 updates OAuth security guidance using deployment experience and stronger protocol requirements. It identifies insecure patterns to stop using and provides a primary reference for reviewing identity integrations.

Reviewed 2026-10-06

Saleor maintainers Maintainer Advisory

Saleor email changes: bind confirmation to account and state

Saleor did not check that an email-change confirmation token belonged to the signed-in account. The maintainer's scenario requires a valid token for one account and authenticated access to another; it describes possible…

Reviewed 2026-10-06

samlify Maintainer Advisory

samlify: signing does not establish claim provenance

samlify could turn user-controlled profile text into extra SAML claims before the identity provider signed them. The maintainer's example shows those added attributes being accepted; gaining privileges would also…

Reviewed 2026-10-03

GitHub Security Lab Research Paper

Sentry: resource ownership must match the authorized organization

Sentry checked a user's event-reading permission without ensuring the requested event belonged to that organization. GitHub Security Lab reports cross-organization disclosure in version 25.12.0 by a signed-in user with…

Reviewed 2026-10-03

SLSA Community Security Standard

SLSA v1.2: supply-chain security and build provenance

SLSA v1.2 defines separate assurance tracks for source code and builds. Its build track moves from recording how an artifact was produced to authenticated hosted builds and stronger isolation, helping consumers check…

Reviewed 2026-10-02

Spree Maintainer Advisory

Spree: cart association must retain guest-possession checks

Spree let a signed-in customer claim an unowned guest cart without checking proof that the cart was theirs. The maintainer describes recoverable ownership and email changes, plus disclosure of checkout addresses if the…

Reviewed 2026-10-03

GitHub Security Lab Research Paper

Spree: guest ownership still requires an authorization proof

Spree treated completed orders with no account owner as readable without the separate guest-order authorization token. GitHub Security Lab found the flaw in version 5.2.6, and the maintainer reports a controlled…

Reviewed 2026-10-03

GitHub Security Lab Research Paper

Sylius: component integrity does not authorize referenced objects

Sylius checked component-state integrity without checking ownership of every object referenced by action arguments. GitHub Security Lab demonstrates cross-customer address disclosure by an authenticated customer in…

Reviewed 2026-10-03

GitHub Security Lab Research Paper

Umbraco: editing an account does not authorize assigning every role

Umbraco checked who could edit an account without consistently checking which roles they could grant. GitHub Security Lab reports administrator membership gained from a non-administrator API account in version 17.2.0.…

Reviewed 2026-10-03

Vikunja Maintainer Advisory

Vikunja: saved favorites must recheck current project access

Vikunja could keep showing favorited tasks after project access was revoked because search did not recheck current permissions. The maintainer reports that updated titles and descriptions remained readable, despite…

Reviewed 2026-10-03

World Wide Web Consortium Technical Standard

Fetch Metadata Request Headers

The Fetch Metadata draft defines browser headers that help servers understand a request's site relationship, destination, mode and user activation. It also explains how redirects and caching affect decisions based on…

Reviewed 2026-10-03

WHATWG Technical Standard

HTML COOP: opener separation and same-origin authority

The HTML standard explains how opener policies separate browser windows during navigation. Breaking the opener connection does not fully isolate same-origin pages, which can still share storage, service workers,…

Reviewed 2026-10-03

GitHub Security Lab Research Paper

Zammad: overridden serialization must preserve group authorization

Zammad's specialized ticket serializer omitted a group-permission check present in its base implementation. GitHub Security Lab reports that an authenticated agent-role user could read confidential tickets and…

Reviewed 2026-10-03

Visual models

A limited recovery request creates an account-bound challenge. Only valid, unexpired, unused proof permits a reset. Failure preserves account state; success consumes the challenge and notifies the owner.

Conceptual model Diagram

Recovery must preserve account ownership

Editorial conceptual model derived from the linked cases and official guidance; not a vendor architecture diagram or an exploitation sequence.

Reviewed 2026-10-02
User intent and untrusted connector content enter the AI workflow separately. A proposed action reaches an independent policy decision. Only an authorized state change proceeds; otherwise state remains unchanged.

Conceptual model Diagram

Retrieved content is data, not authority

Editorial conceptual model derived from the linked cases and official guidance; not a vendor architecture diagram or an exploitation sequence.

Reviewed 2026-10-02
An incoming browser message first passes origin, sender-context and format validation. A separate decision checks the operation and recipient. Failed checks reject the message without disclosure or state change. Approved content remains data and only the permitted action is performed.

Conceptual model Diagram

Browser messages need separate trust checks

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.

Reviewed 2026-10-02
Untrusted contributions remain separated from trusted build state. An artifact and its provenance reach a consumer policy gate. Only authenticated evidence matching the artifact and expected builder and inputs makes it eligible for release review; missing or mismatched evidence prevents promotion.

Conceptual model Diagram

Build evidence must match the artifact and trusted builder

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…

Reviewed 2026-10-03
A caller requests a combined view using required source A and required source B. Each source has its own authorization decision for that caller. Either No denies the combined view without protected source data. Both Yes decisions are required at an explicit AND gate before composition. Field selection preserves each source’s field permissions, and only permitted data is returned.

Conceptual model Diagram

Combined views preserve every source's access boundary

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…

Reviewed 2026-10-04
Approved grant context, including subject, client, permitted resources and actions, and a first-issuance or renewal request are both required inputs. Resolve trusted grant context and check that the grant and client are correct and requested authority remains within approval. Issue only bounded authority when all checks pass. Do not issue on mismatch or incomplete binding. Changing the approved authority requires a separate authorization decision; no automatic-consent loop is shown.

Conceptual model Diagram

Delegated authority stays within the approved grant

Original editorial defensive synthesis, not a vendor architecture, protocol sequence or independently verified patch. The Google researcher describes lost client and permission binding through a device grant; proposed…

Reviewed 2026-10-04
A failure retains the original caller, action and resource. An explicit access denial stops without protected data. A known recoverable operational failure can propose a fallback, but switching execution identity grants no extra caller entitlement. A separate application-policy decision evaluates that fallback for the original caller. Denied or indeterminate decisions stop without protected data. A permitted, scoped fallback returns only caller-permitted data.

Conceptual model Diagram

Fallbacks must preserve the original caller's authority

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…

Reviewed 2026-10-04
The identity provider authenticates a subject and binds issued claims to it. The relying application validates the issuer, audience, signature, and ownership binding before mapping to a local account. Failed validation is rejected.

Conceptual model Diagram

An identity claim must belong to the user

Editorial conceptual model derived from the linked cases and official guidance; not a vendor architecture diagram or an exploitation sequence.

Reviewed 2026-10-02
A request enters a server-side caller, resource and operation authorization decision. Denial returns no protected data. Approval proceeds to explicit field selection before serialization. Only permitted, necessary fields cross into client-visible data. A separate consumer-context handling step keeps content, including generated text, from acquiring executable meaning before display. Browser rendering never supplies server authorization.

Conceptual model Diagram

Server disclosure and browser interpretation

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…

Reviewed 2026-10-03
A verified workload identity and a requested operation enter an independent authorization decision. Policy checks the role, action, resource and tenant together. Only the approved resource scope is allowed; other requests are denied. Both decisions produce an audit record.

Conceptual model Diagram

Keep workload authority tenant-scoped

Editorial conceptual model derived from the linked cases and official guidance; not a vendor architecture diagram or an exploitation sequence.

Reviewed 2026-10-02

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.

Library synced 2026-11-18 · cases last reviewed 2026-10-05

af2809c308fc · JSON exports & schemas · CC BY 4.0 content / MIT software