Broken object-level authorization in Elastic Kibana's Fleet module exposes sensitive proxy configuration data to under-privileged users who should not have access to it. A user holding only the 'read agent policies' permission - a scope far narrower than Fleet administrative access - can read the full Fleet proxy configuration, including proxy authentication credentials and private key material. No public exploit code has been identified at time of analysis, but the CVSS Scope Changed metric (S:C) reflects that stolen credentials could be leveraged against proxy infrastructure well beyond the Kibana instance itself.
Out-of-bounds read in IBM i 7.3 through 7.6 enables a remote authenticated attacker to read beyond allocated memory boundaries, leaking sensitive information and causing limited availability impact. The CVSS vector (AV:N/AC:L/PR:L/UI:N) confirms network-exploitable attack requiring only low-privilege credentials with no user interaction. No public exploit code and no CISA KEV listing have been identified at time of analysis, but the patch is available via IBM's support portal.
Denial of service in Elasticsearch's wildcard matching component allows any authenticated user with search access to terminate an affected node by submitting a single crafted query. The recursive wildcard evaluator has no bounds on recursion depth or total match operations; a pattern containing many wildcard groups evaluated against a sufficiently long field name exhausts the JVM thread stack, triggering a StackOverflowError that Elasticsearch treats as unrecoverable, causing immediate node shutdown rather than graceful request failure. No public exploit has been identified at time of analysis, but exploitation requires only low-privilege search access (CVSS PR:L), meaning any application service account or authenticated API user represents a viable threat origin.
Denial-of-service in Kibana's connector reporting endpoint allows any authenticated user - including those with read-only privileges - to permanently incapacitate the Kibana process with a single crafted request. The flaw stems from accepting arbitrarily large query expressions without size limits, causing the process to consume unbounded CPU evaluating the expression and rendering the instance unable to serve any user until manually restarted. No public exploit has been identified, but the low privilege bar (read-only access) and single-request reproducibility make this an attractive target for internal threat actors or compromised low-privilege accounts seeking to disrupt SOC or observability operations.
Memory exhaustion in Elasticsearch's search highlighting subsystem allows any authenticated read-only user to terminate a node process with a single crafted request. The highlighting API accepts an unbounded user-supplied count, and the internal data structure allocated from that count is not tracked by Elasticsearch's circuit breaker system, bypassing the existing OOM safeguards entirely. The defect is non-volumetric - one small request is sufficient - and degrades cluster routing and health beyond the targeted node. No public exploit has been identified at time of analysis, though the low attack complexity and minimal privilege requirement make this a realistic internal threat.
Uncontrolled recursion in Elasticsearch's custom analysis definition processing allows an authenticated low-privileged user to crash an individual cluster node with a single request. By submitting a malformed index creation request whose custom analyzer, tokenizer, or filter components reference each other without termination, the resolution logic enters unbounded recursion, exhausts the JVM thread stack, and terminates the targeted node. No public exploit has been identified at time of analysis, but the attack requires only index creation privileges - a permission commonly granted to application service accounts - making the practical barrier lower than the authentication prerequisite might suggest.
Uncontrolled recursion in Elasticsearch's search request processing allows an authenticated user with only read-level privileges on a single index to crash any cluster node by submitting a specially crafted query containing pathologically nested structures. The absence of a recursion depth limit causes thread stack exhaustion during deserialization of the nested payload, triggering an unhandled stack overflow that terminates the affected node. No public exploit code has been identified at time of analysis, and CISA KEV does not list this CVE, but the low privilege bar (read-only access) makes the attack surface broad in multi-tenant or exposed deployments.
Elasticsearch exposes an authenticated denial-of-service path where a user holding only read privileges on a single index can crash an entire cluster node with one small, crafted search request. Rooted in CWE-789 (Memory Allocation with Excessive Size Value) and matching CAPEC-130 (Excessive Allocation), the flaw causes the JVM heap to be exhausted, terminating the affected node and disrupting all workloads on it. No CISA KEV listing and no public exploit code have been identified at time of analysis, but the low attack complexity makes this a credible threat in multi-tenant or broadly-credentialed Elasticsearch deployments.
ES|QL query processing in Elasticsearch allows an authenticated low-privileged user to exhaust heap memory on any receiving cluster node by submitting a specially crafted query, rendering that node unavailable. The CVSS 6.5 score (AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H) reflects remote, low-complexity exploitation requiring only a valid account, with high availability impact but no data exposure. Elastic has released fixes in versions 8.18.0 and 9.0.0 per advisory ESA-2026-111; no public exploit or active exploitation has been identified at time of analysis.
Resource exhaustion in Elastic Kibana allows an authenticated user who holds maintenance window management privileges to crash the Kibana process for all users by submitting a specially crafted malformed payload. The Kibana process consumes unbounded resources and becomes fully unresponsive, requiring manual operator intervention to restore service - there is no automatic recovery. No public exploit code has been identified and this vulnerability is not listed in the CISA KEV catalog; however, the low access complexity and broad multi-tenant availability impact make it a meaningful operational risk for shared Kibana deployments.
Denial of service in Elastic Kibana's alerting feature allows an authenticated user with read-only alerting privileges to crash the Kibana process for all users by submitting a single crafted malformed payload. The resource exhaustion is unbounded - a single HTTP request is sufficient to render Kibana unable to serve any users until the process is manually restarted. No public exploit code has been identified at time of analysis, and this vulnerability is not listed in the CISA KEV catalog; however, the low attack complexity and broad impact make it a meaningful risk in any multi-tenant Kibana deployment where alerting access is granted to partially-trusted users.
Algorithmic complexity exhaustion in Kibana's TSVB (Time Series Visual Builder) expression evaluator allows any authenticated user to render the entire Kibana instance unresponsive with a single crafted request. The synchronous evaluation of a specially crafted deeply nested expression monopolizes the Kibana request-processing thread indefinitely, blocking all subsequent requests until the service is manually restarted. Patched versions 8.19.20 and 9.4.5 are available per Elastic advisory ESA-2026-104; no public exploit or CISA KEV listing has been identified at time of analysis.
Denial of service in Elastic Kibana allows any authenticated low-privileged user to crash the entire Kibana service for all users by submitting a specially crafted, malformed payload to a visualization feature. The root cause is CWE-770 (resource allocation without throttling): Kibana fails to validate the malformed input before processing, causing unbounded heap growth until the host OS kills the process. No public exploit has been identified at time of analysis, and exploitation requires authenticated access, but the low privilege bar and total service disruption make this a meaningful insider or compromised-credential threat. Patch versions 8.19.20 and 9.4.5 are referenced in the vendor advisory ESA-2026-100.
Unbounded resource allocation in Kibana's Observability log analysis validation feature enables any authenticated low-privilege user to exhaust the Kibana process memory and deny service to all users. The flaw (CWE-770, CAPEC-130) requires only minimal credentials - no elevated roles - and a single inexpensive HTTP request triggers concurrent work without throttling. Notably, the request can be repeated trivially, sustaining pressure even on well-provisioned deployments that might otherwise degrade gracefully rather than crash outright. No public exploit code or active exploitation (CISA KEV) has been identified at the time of analysis, though the low exploitation bar makes this a credible insider or credential-abuse threat.
Denial of service in Kibana's Playground for RAG feature allows a low-privileged authenticated user to exhaust the memory and processing resources of a Kibana instance via a single crafted request. The vulnerable endpoint accepts a user-supplied list of document fields without enforcing any length bound or deduplication, enabling a request to trigger construction of a response orders of magnitude larger than the underlying data. No active exploitation has been confirmed (not in CISA KEV), and no public exploit code has been identified at time of analysis. Elastic has released fixes in versions 9.3.8 and 9.4.4.
Denial of service in Elasticsearch allows any authenticated user holding only read-only index privileges to crash a cluster node by submitting a single crafted search request. The intervals query type does not enforce Elasticsearch's configurable input-length restriction, so a deeply nested pattern drives unbounded recursion that exhausts the JVM thread stack and terminates the node process. No public exploit or CISA KEV listing exists at time of analysis; however, the low complexity and minimal privilege bar make this a credible internal threat, and Elastic has released fixes in versions 8.19.20 and 9.4.5 per ESA-2026-81.
Memory exhaustion in Elasticsearch allows any authenticated user with only read privileges to crash an Elasticsearch node with a single small crafted API request. The root cause is missing validation of a user-controlled size value that is passed directly to an internal memory reservation routine, enabling uncontrolled heap allocation that terminates the node process and degrades cluster health. No public exploit code has been identified at time of analysis, and the vulnerability is not listed in CISA KEV, but the non-volumetric nature - one request is sufficient regardless of heap configuration - significantly lowers the bar for disruption by low-privileged insiders or compromised service accounts.
Denial of service in Elasticsearch allows any low-privileged authenticated user to crash an affected cluster node with a single crafted HTTP request. The root flaw is a validate-after-deserialize ordering error: Elasticsearch decodes and deserializes an opaque identifier before confirming the cluster issued it, and an attacker-controlled size value embedded in the forged identifier drives an uncapped heap allocation (CWE-789) that exhausts available memory and fatally terminates the node process. No public exploit code has been identified at time of analysis, and this CVE is not listed in the CISA KEV catalog.
Denial of service in Elasticsearch allows an authenticated user with only read privileges to crash a cluster node by submitting a crafted search request. Processing the malicious input triggers unbounded memory allocation in a specific internal component, bypassing existing memory accounting controls and causing a fatal out-of-memory condition that terminates the affected node process. No public exploit code has been identified at time of analysis, but the low privilege requirement (read-only access) significantly broadens the potential attacker pool in multi-tenant or SaaS deployments.
Stack-based buffer overflow in IBM i 7.3 through 7.6 allows a remote attacker with low-privileged network access to crash the system, causing a denial of service. The CVSS vector (AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H) confirms network reachability at low complexity, but impact is limited strictly to availability - no confidentiality or integrity consequences are indicated. IBM has released a patch; no public exploit or active exploitation has been identified at time of analysis.
Broken Access Control in the Motors WordPress plugin (car dealership and classified listings, by Stylemix) through version 1.4.113 allows subscriber-level authenticated users to perform privileged actions that should be restricted to higher roles, resulting in a high integrity impact. The flaw is classified as CWE-862 (Missing Authorization), meaning the plugin fails to verify whether the requesting user has sufficient privileges before executing sensitive operations. No public exploit has been identified at time of analysis, and the vulnerability is not listed in the CISA KEV catalog.
Cleartext token exposure in Elastic Cloud on Kubernetes allows any principal with namespace workload-spec read access to retrieve live Elasticsearch service account credentials. Affected deployments are those running Fleet Server resources configured to authenticate to Elasticsearch via service account tokens - ECK writes the token directly into the generated Pod specification in cleartext rather than referencing the Kubernetes Secret where other credentials are stored, bypassing an intentional RBAC boundary. No public exploit has been identified and the vulnerability is not listed in CISA KEV, but the confidentiality impact is high given that the exposed credential grants direct Elasticsearch data-plane access.
Sensitive customer data exposure in the WP Event Solution WordPress plugin (versions up to and including 4.1.18) allows authenticated low-privileged users to access confidential customer information over the network. The vulnerability, classified under CWE-201 (Insertion of Sensitive Information Into Sent Data), results in high confidentiality impact with no integrity or availability loss. No public exploit code or CISA KEV listing has been identified at time of analysis, but the low attack complexity and network accessibility elevate realistic risk for sites collecting customer data via this plugin.
Policy information disclosure in Elastic Fleet Server allows any enrolled agent with valid credentials to retrieve policy documents it was not assigned to by substituting a client-controlled key in artifact download requests. The authorization decision relied on a client-supplied value that was persisted without being cross-checked against the server-side record of the requesting agent's actual policy assignment - a textbook CWE-639 flaw. The CVSS 3.1 score of 6.5 (AV:N/AC:L/PR:L/UI:N) reflects network-accessible, low-complexity exploitation by any authenticated agent; no public exploit has been identified at time of analysis, and the vulnerability is not listed in CISA KEV.
Missing authorization in Elastic Kibana's Security solution allows users holding only detection rule authoring privileges to associate and indirectly trigger Elastic Defend endpoint response actions - including host isolation, process termination, and remote execution - against managed hosts, without ever being granted the dedicated Endpoint response action privileges that are supposed to govern those capabilities. The exploitation path is indirect: a restricted user crafts a detection rule with embedded response actions, and when that rule generates alerts against matching endpoints, the privileged actions execute autonomously through the Elastic Defend agent. No public exploit has been identified at time of analysis, and the flaw is not listed in the CISA KEV catalog.
Path traversal in Zimbra Collaboration Suite (ZCS) Briefcase document editing exposes sensitive web application files to authenticated low-privileged users. The vulnerability stems from improper validation of the `packages` parameter, enabling directory escape via crafted traversal sequences. No public exploit code or CISA KEV listing has been identified at time of analysis; the vendor has released a fix in version 10.1.17.
Missing authorization in Kibana's internal data retrieval capability for Elastic Defend endpoint response actions exposes sensitive security telemetry to any authenticated low-privileged user. The internal API endpoint used for endpoint response actions bypasses the Security Solution and endpoint privilege checks applied to its user-facing equivalents, executing with elevated internal permissions rather than the requesting user's permissions. No public exploit or CISA KEV listing exists at time of analysis; however, any authenticated Kibana user - regardless of Security Solution role assignments - can exploit this to read endpoint response records and host-returned response content from managed endpoints.
XML External Entity (XXE) injection in IBM i 7.3 through 7.6 allows remote authenticated attackers to exfiltrate sensitive information by submitting crafted XML payloads to an XML-processing component on the platform. The CVSS vector (AV:N/AC:L/PR:L) confirms low-complexity network exploitation requiring only a low-privileged account, with high confidentiality impact and no integrity or availability effect. No active exploitation has been confirmed (not in CISA KEV), and IBM has released a patch via its support portal.
Cross-namespace secret exfiltration in the Elastic Cloud on Kubernetes (ECK) operator allows any low-privileged Kubernetes tenant to read secrets from arbitrary namespaces cluster-wide. The operator's reconciliation loop blindly trusts namespace references embedded in secret annotations without verifying whether the requesting user is authorized to access the referenced namespace, enabling a classic confused-deputy (CWE-441) privilege escalation that breaks Kubernetes RBAC namespace isolation. No public exploit code has been identified at time of analysis, but the low attack complexity and high confidentiality impact make this a material priority for any multi-tenant Kubernetes deployment running ECK; vendor-released patch version 3.5.0 is available per ESA-2026-121.
Broken access control in Contact Form 7 - PayPal & Stripe Add-on versions 2.5.1 and below allows unauthenticated remote attackers to invoke privileged plugin functionality without authorization, enabling unauthorized modification of plugin state or disruption of payment form behavior. Discovered and disclosed by Patchstack (ENISA EUVD-2026-57898), the flaw is rooted in CWE-862 (Missing Authorization), meaning one or more plugin endpoints or AJAX handlers execute sensitive actions without verifying caller permissions. No public exploit code has been identified at time of analysis, and the vulnerability is not listed in the CISA Known Exploited Vulnerabilities catalog.
Broken access control in the AI for SEO WordPress plugin (versions ≤ 2.4.2) permits unauthenticated remote attackers to invoke restricted plugin functionality without credentials. The CVSS vector (AV:N/AC:L/PR:N/UI:N) confirms exploitation requires no authentication, no user interaction, and no special network position - any internet-facing WordPress site running the affected plugin is reachable. Impact is bounded to partial integrity and availability disruption (I:L/A:L) with no confidentiality exposure, suggesting unauthorized manipulation of plugin operations rather than data exfiltration. No public exploit identified at time of analysis, and the vulnerability is not listed in CISA KEV.
Broken access control in WP Social Avatar WordPress plugin (versions ≤ 1.5) exposes one or more plugin endpoints to unauthenticated remote attackers, who can perform restricted operations without valid credentials. The CVSS vector (AV:N/AC:L/PR:N/UI:N) confirms network-accessible exploitation with no credentials or user interaction required, with consequences limited to low-severity integrity and availability impact and no confidentiality exposure. No public exploit code has been identified at time of analysis, and the vulnerability does not appear in the CISA Known Exploited Vulnerabilities catalog.
Broken access control in the Secure Card Gateway for ePay Paycenter WordPress plugin (versions ≤ 1.0.32) enables remote unauthenticated attackers to bypass authorization checks and invoke privileged plugin operations without supplying any credential. Reported by Patchstack and classified as CWE-862 (Missing Authorization), the flaw carries a CVSS 3.1 score of 6.5 with a fully network-accessible, zero-privilege, zero-interaction attack vector. No active exploitation has been confirmed - this CVE does not appear in the CISA KEV catalog - and no public proof-of-concept code has been identified at time of analysis.
Unauthenticated IDOR in the Do Lasso WordPress affiliate link management plugin (versions 358 and earlier) allows remote attackers without any credentials to reference and manipulate internal objects by supplying attacker-controlled identifier values, resulting in unauthorized modification or disruption of affiliate link data. The CVSS vector (AV:N/AC:L/PR:N/UI:N) confirms exploitation requires no authentication and no user interaction against default plugin installations. No public exploit code or CISA KEV listing has been identified at time of analysis, though the low attack complexity and zero-privilege requirement lower the barrier to opportunistic abuse.
Memory exhaustion and unbounded disk consumption in Mattermost's Boards archive import handler expose self-hosted deployments to authenticated denial-of-service. Affected versions span the 10.11.x, 11.7.x, 11.8.x, and 11.9.x release trains, all of which fail to enforce decompressed content size limits or respect the configured maximum file size when processing .boardarchive uploads. Any authenticated user with access to the Boards import endpoint can trigger this condition; no public exploit has been identified at time of analysis and the vulnerability does not appear in CISA KEV.
Privilege escalation in Kibana Fleet allows authenticated low-privileged users to expand the write scope of Elasticsearch API keys issued to Elastic Agents beyond intended boundaries. By declaring extra data streams in an integration policy, an attacker causes Fleet to over-provision the API key given to enrolled agents, granting write-insert and index-mapping-extension privileges on indices outside the legitimate scope of that integration. No public exploit has been identified at time of analysis, and the flaw has not been listed in the CISA KEV catalog, but the network-accessible attack vector and low privilege requirement make this a meaningful integrity risk in multi-tenant or shared Elastic Stack deployments.
Stored Cross-Site Scripting in the Accordion WordPress plugin (versions <= 3.0.6) allows authenticated subscriber-level users to inject malicious JavaScript into accordion content fields, which then executes in the browser of any higher-privileged user who views the affected page. The scope change (S:C in CVSS) confirms the attack pivots from the subscriber's input to the victim's browsing session, enabling session hijacking, credential theft, or unauthorized admin actions. No public exploit code or CISA KEV listing is identified at time of analysis, but the low attack complexity and subscriber-level access requirement make this a realistic threat on sites with open user registration.
Subscriber-level Cross-Site Scripting in FluentCommunity (WordPress plugin) versions up to and including 2.7.5 allows authenticated users with subscriber-level access to inject malicious scripts that execute in the browser context of other users, including administrators. The scope change (S:C in CVSS) indicates the injected payload affects a different security principal than the attacker, enabling session hijacking, credential theft, or unauthorized administrative actions. No public exploit code or CISA KEV listing has been identified at time of analysis, though the low privilege requirement and network-accessible attack surface make this a realistic threat for multi-user WordPress deployments running community features.
Cross-site scripting in the AfterShip Tracking WordPress plugin (versions ≤ 1.18.1) enables subscriber-level authenticated users to inject malicious scripts that execute in the browsers of other site visitors or administrators. The CVSS scope change (S:C) confirms the injected payload executes outside the attacker's own session context, making administrator session hijacking the primary impact. No public exploit code or active exploitation via CISA KEV has been identified at time of analysis.
Stored cross-site scripting in the Profile Extra Fields by BestWebSoft WordPress plugin (versions ≤1.3.4) allows an authenticated subscriber-level user to inject arbitrary JavaScript into custom profile fields, which then executes in the browser of any higher-privileged user who views the affected profile - including site administrators. The scope-changed impact (S:C) is the key risk multiplier: a minimally privileged attacker can potentially hijack admin sessions or perform unauthorized actions on behalf of victims. No public exploit code or active exploitation has been identified at time of analysis.
Subscriber-level cross-site scripting in AcyMailing SMTP Newsletter plugin (versions up to and including 10.11.1) enables a low-privileged authenticated user to inject malicious script payloads that execute in the browser context of higher-privileged site users, such as administrators. The changed-scope CVSS vector (S:C, PR:L, UI:R) is consistent with stored XSS where subscriber-submitted content persists and fires when an administrator reviews it in the WordPress backend, enabling session hijacking or unauthorized administrative actions. No public exploit code and no active exploitation have been identified at time of analysis.
Local file inclusion in Zimbra Collaboration's Classic Web Client (versions before 10.1.17) exposes protected server-side files via the Forward servlet's unvalidated 'fu' request parameter. An attacker supplying a crafted path can read files such as WEB-INF/web.xml, leaking application configuration including database credentials, internal class mappings, and security settings. The vulnerability carries a low CVSS score of 3.1, but the confidentiality impact of web.xml disclosure can be disproportionately severe since that file often contains secrets used to pivot further. No public exploit code or CISA KEV listing has been identified at time of analysis.
The iscsiuio DHCP parser in open-iscsi crashes when processing maliciously crafted DHCP reply packets, disrupting iSCSI storage connectivity or network boot on affected hosts. An integer underflow (CWE-191) in the IPv4/UDP DHCP reply parsing path triggers an out-of-bounds read, causing the iscsiuio process to terminate; impact is limited to availability with no confidentiality or integrity exposure. No public exploit code or active exploitation has been identified at time of analysis, and the adjacent-network attack vector constrains the realistic attacker population to the same Layer-2 segment as the target.
Broken access control in Internal Link Optimiser (WordPress plugin by Toast Plugins) versions up to and including 5.2.7 exposes restricted plugin functionality to unauthenticated network attackers. The CVSS vector (AV:N/AC:L/PR:N/UI:N) confirms that no credentials or user interaction are required, allowing any remote party to interact with privileged plugin endpoints on affected WordPress installations. The impact is limited to partial confidentiality and integrity compromise - consistent with unauthorized read or write access to internal link data - with no availability impact identified. No public exploit code or CISA KEV listing has been identified at time of analysis.
JWT signing secret forgery in Zimbra Collaboration before 10.1.17 stems from the zimbraDocumentEditingJwtSecret being generated with an insecure pseudo-random number generator, yielding insufficient entropy. An authenticated attacker who captures a JWT token from the OnlyOffice integration can perform offline brute-force to recover the signing secret, then mint arbitrary forged tokens - enabling impersonation or unauthorized document access within the OnlyOffice workflow. No public exploit code and no CISA KEV listing have been identified at time of analysis; exploitation requires multi-step conditions including a valid Zimbra account and captured JWT traffic.
Stored Cross-Site Scripting in GiveWP's donor data handling allows a low-privileged attacker to inject malicious JavaScript into donor-submitted fields that executes in a privileged user's browser when donation records are viewed. All GiveWP installations running versions prior to 4.16.6 are affected. The CVSS scope change (S:C) indicates the injected script can operate outside the donor submission context - most likely executing within an administrator's session, enabling session hijacking or unauthorized admin actions. No public exploit or CISA KEV listing has been identified at time of analysis.
Stored Cross-Site Scripting in the Featured Image from URL WordPress plugin (versions up to and including 5.3.3) allows authenticated Contributors to inject malicious scripts into pages viewed by higher-privileged users such as editors or administrators. The vulnerability resides in insufficient output encoding of user-supplied URL input used for featured images, giving attackers a foothold for privilege escalation or session hijacking within the WordPress admin context. No public exploit code or CISA KEV listing has been identified at time of analysis, but the Contributor-level access requirement makes this realistic in any multi-author WordPress site.
Stored Cross-Site Scripting in WpBookingly (WordPress plugin by MagePeople Team, versions ≤ 1.3.2) allows a low-privileged attacker - such as an external customer submitting a booking request - to inject malicious scripts into customer-facing input fields that are later rendered unsanitized in the WordPress admin panel. When an administrator views the poisoned booking data, the injected script executes in their browser context, enabling session hijacking, credential theft, or unauthorized admin-level actions. No public exploit code or CISA KEV listing has been identified at time of analysis, but the network-accessible, low-complexity attack surface makes this a meaningful risk for any site accepting public booking submissions.
Cross-site scripting in the Popup by Supsystic WordPress plugin (all versions up to and including 1.11.2) enables script injection into pages viewed by other users, with the injected code executing in the victim's browser under a changed scope. Reported by Patchstack and tracked as EUVD-2026-57935, the vulnerability carries a CVSS base score of 6.5 with Changed Scope, meaning successful exploitation crosses the security boundary between the plugin and the victim's browser session. No public exploit code or CISA KEV listing has been identified at time of analysis, but the low attack complexity and network-accessible attack vector make it straightforward to weaponize given the right conditions.
Privilege escalation in Gitea Open Source Git Server before 1.25.5 allows an attacker holding a leaked read-only API token to mint a runner registration token and register a malicious Actions runner against the instance. The root cause is that the runner registration-token endpoints - at the user, organization, repository, and admin levels - are classified as read-only by the API scope middleware solely because they use the HTTP GET method, despite creating a token if none exists. With a valid registration token, the attacker can register a self-hosted runner that executes CI/CD workflow jobs, gaining access to repository secrets and source code. No public exploit code has been identified at time of analysis, and the EPSS score of 0.16% (6th percentile) indicates low observed exploitation probability.