Vulnerabilities
All
My stack
Known exploited
Public exploit
Critical (9+)
Likely exploited (EPSS ≥ 50%)
Changed, 24 h
New, 24 h
Sort:
risk
modified
published
CVSS
EPSS
| CVE | CVSS | EPSS | Vendor, product | Description | Modified |
|---|---|---|---|---|---|
| CVE-2026-78633 | 6.5 medium | 0.00% | no CPE yet | The Okta RADIUS Agent does not validate the sub-attribute length when parsing a Vendor-Specific attribute. A sub-attribute length of zero causes the parser to loop indefinitely, resulting in worker-thread exhaustion and denial of new RADIUS authentication requests. | 16 Oct |
| CVE-2026-77364 | 6.5 medium | 0.00% | no CPE yet | obsidian-tc is a governed, agent-ready Model Context Protocol server for Obsidian vaults. Prior to 1.3.1, undefined readPaths and writePaths settings allow all paths, strictReadDefault does not protect direct requests, and the .obsidian, .git, and .trash control directories are excluded only during enumeration. An authenticated caller can directly read sensitive files such as .obsidian/plugins/*/data.json, potentially exposing plugin API keys, Obsidian Sync passwords, and other in-vault data. Version 1.3.1 fixes the vulnerability by denying the control directories by default and enforcing strictReadDefault on direct request paths. | 16 Oct |
| CVE-2026-101335 | 6.5 medium | 0.00% | no CPE yet | The Forminator Forms – Contact Form, Payment Form & Custom Form Builder plugin for WordPress is vulnerable to Arbitrary Shortcode Execution in versions up to, and including, 1.58.1. This is due to the forminator_replace_quiz_form_data() helper invoking do_shortcode() on the fully expanded notification body once the {quiz_result} merge tag has been substituted, while the earlier-substituted {user_agent} and {refer_url} merge tags carry unauthenticated attacker-controlled bytes (the User-Agent header and the referer_url POST parameter) through sanitize_text_field()/esc_html(), neither of which strip the '[' and ']' shortcode delimiters. This makes it possible for unauthenticated attackers to execute arbitrary WordPress shortcodes registered on the site by submitting a quiz whose notification template contains {quiz_result}; the resulting impact depends on the shortcodes exposed by other installed plugins/themes and can range from arbitrary content injection to information disclosure. | 16 Oct |
| CVE-2026-100394 | 6.5 medium | 0.00% | no CPE yet | Nezha Monitoring is a self-hostable, lightweight, servers and websites monitoring and O&M tool. From 2.2.7 until 2.2.11, ServiceSentinel.worker in service/singleton/servicesentinel.go checked an owned service under servicesLock and later dereferenced pointer-valued serviceStatusToday and serviceCurrentStatusData entries under serviceResponseDataStoreLock without revalidating that ServiceSentinel.Delete had not removed them. A RoleMember could concurrently report TCP or ICMP results through the member's own agent and delete the same owned service, causing a nil pointer dereference. Because the gRPC server had no recovery interceptor, the worker panic terminated the whole dashboard process and disrupted every tenant's web interface, REST API, and agent connections. The issue affects availability only and does not disclose or modify data. This issue is fixed in versions 2.2.11 and 2.3.0. | 14 Oct |
| CVE-2026-107287 | 6.5 medium | 0.00% | pydantic, pydantic_ai | Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 1.77.0 until 1.107.7 and 2.52.0, the local web_fetch_tool and the WebFetch local fallback can consume excessive CPU and memory during HTML-to-Markdown conversion of attacker-controlled HTML containing deeply nested block elements. Conversion repeatedly reprocesses accumulated text and can greatly expand intermediate output before the returned-content limit is applied, allowing a model-directed fetch to delay other work in the process. Provider-native web fetching is not affected. This issue is fixed in versions 1.107.7 and 2.52.0. | 14 Oct |
| CVE-2026-107294 | 6.5 medium | 0.00% | pydantic, pydantic_ai | Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 1.77.0 until 1.107.2 and 2.24.0, web_fetch_tool, the WebFetch local fallback, and remote FileUrl media downloads buffer the complete HTTP response body before enforcing content-size controls. An attacker-influenced URL can stream an arbitrarily large response that exhausts process memory and crashes the worker; affected media types include ImageUrl, DocumentUrl, VideoUrl, and AudioUrl. SSRF protections remain effective, and the impact is limited to availability. This issue is fixed in versions 1.107.2 and 2.24.0. | 13 Oct |
| CVE-2025-10540 | 6.5 medium | 0.00% | no CPE yet | iMonitor EAM 9.6394 transmits communication between the EAM client agent and the EAM server, as well as between the EAM monitor management software and the server, in plaintext without authentication or encryption. An attacker with network access can intercept sensitive information (such as credentials, keylogger data, and personally identifiable information) and tamper with traffic. This allows both unauthorized disclosure and modification of data, including issuing arbitrary commands to client agents. | 11 Oct |
| CVE-2026-91662 | 6.4 medium | 0.00% | oracle, essbase | Vulnerability in Oracle Essbase (component: Java Based Agent). The supported version that is affected is 21.8.3.0.0. Difficult to exploit vulnerability allows low privileged attacker with logon to the infrastructure where Oracle Essbase executes to compromise Oracle Essbase. While the vulnerability is in Oracle Essbase, attacks may significantly impact additional products (scope change). Successful attacks of this vulnerability can result in unauthorized access to critical data or complete access to all Oracle Essbase accessible data as well as unauthorized update, insert or delete access to some of Oracle Essbase accessible data. CVSS 3.1 Base Score 6.4 (Confidentiality and Integrity impacts). CVSS Vector: (CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:C/C:H/I:L/A:N). | Tue 27 Oct |
| CVE-2026-13318 | 6.4 medium | 0.00% | kubevirt, redhat, openshift_virtualization | A server-side request forgery (SSRF) flaw was found in KubeVirt's virt-api port-forward handler. When processing a port-forward request to a VirtualMachineInstance (VMI), virt-api reads the target IP from vmi.Status.Interfaces[0].IP and passes it directly to net.Dial() without validation. For VMIs using non-masquerade network bindings (bridge or secondary-only), this IP is reported by the QEMU guest agent running inside the VM and is fully controllable by the VM owner. An attacker with kubevirt.io:edit permissions can create a VM with a modified guest agent that reports an arbitrary IP address, then request port-forward to establish a bidirectional TCP tunnel from virt-api's cluster-internal network position to any routable destination, bypassing NetworkPolicy isolation. | Sun 25 Oct |
| CVE-2026-75130 | 6.4 medium | 0.00% | no CPE yet | Context7 through 2.1.2 contains a prompt injection vulnerability that allows attackers to execute malicious instructions in connected AI coding agents by injecting unsanitized content through the Custom AI Instructions feature served via the MCP server. Attackers can poison the custom instructions to exfiltrate credentials from environment files to an attacker-controlled service and perform destructive file deletion on the victim's machine when the agent makes a routine library documentation request. | 22 Oct |
| CVE-2026-93908 | 6.4 medium | 0.00% | no CPE yet | The Real Estate Manager – Property Listing and Agent Management plugin for WordPress is vulnerable to Stored Cross-Site Scripting via the 'before_price_text' parameter in all versions up to, and including, 7.3 due to insufficient input sanitization and output escaping. This makes it possible for authenticated attackers, with subscriber-level access and above, to inject arbitrary web scripts in pages that will execute whenever a user accesses an injected page. The vulnerability is further enabled by the absence of any capability, nonce, or ownership check on the wp_ajax_rem_create_pro_ajax handler, and because the value is persisted via update_post_meta rather than post_content, the wp_kses filtering tied to the unfiltered_html capability does not apply. | 21 Oct |
| CVE-2026-12254 | 6.4 medium | 0.00% | no CPE yet | Netskope is notified about a potential gap in its agent (NSClient) where a system with Netskope Client while connecting to an access point with Captive Portal switches to a Network which already has internet connectivity, the Captive Portal does not close. This allows the possibility of abuse via the Captive Portal using a rogue access point to either access malicious sites or exfiltrate data.The issue impacts all versions prior to R142 | 20 Oct |
| CVE-2026-107292 | 6.4 medium | 0.00% | pydantic, pydantic_ai | Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 1.34.0 until 2.30.0, the Agent.to_web() and clai web development chat server does not validate the Host header, allowing a website visited by a developer to use DNS rebinding to reach a loopback-hosted agent as a same-origin service. The hostile page can read the served UI and submit chat requests that execute agent tools with the local process's privileges and credentials, causing data disclosure or unwanted side effects. Binding to localhost, Origin checks, and CSRF tokens do not prevent the same-origin DNS rebinding path. This issue is fixed in versions 1.107.5 and 2.30.0. | 14 Oct |
| CVE-2026-116575 | 6.3 medium | 0.00% | no CPE yet | Mem0 provides persistent memory infrastructure for AI agents and applications. Prior to Python SDK 2.0.18 and TypeScript SDK 3.1.6, _build_session_scope() in mem0/memory/main.py and Memory.buildSessionScope() in the TypeScript SDK join caller-controlled user_id, agent_id, and run_id values using unescaped & and = delimiters. A caller can embed another scope component in a controlled identifier so that distinct tenant tuples produce the same session_scope key. Memory.add(), AsyncMemory.add(), and the REST POST /memories flow use that key to read and save recent raw messages, allowing cross-tenant conversation disclosure and eviction or interleaving of another tenant's history. This issue is fixed in Python SDK 2.0.18 and TypeScript SDK 3.1.6. | Tue 27 Oct |
| CVE-2026-6420 | 6.3 medium | 0.00% | no CPE yet | A flaw was found in Keylime. An attacker with root access on an enrolled monitored machine, where the Keylime agent runs, can exploit a vulnerability in the Keylime verifier. The verifier uses a hardcoded challenge nonce for Trusted Platform Module (TPM) quote attestation instead of a cryptographically random value. This allows the attacker to stockpile valid TPM quotes and replay them to evade detection after compromising the system. This issue affects only the push model deployment. | Mon 26 Oct |
| CVE-2026-53851 | 6.3 medium | 0.00% | openclaw | OpenClaw before 2026.5.12 contains a notification bypass vulnerability allowing Slack reaction events to enter the agent pipeline despite disabled reaction notifications. Attackers can trigger unintended agent processing by sending reaction events when the feature is enabled, potentially leading to unauthorized processing of lower-trust input. | Sun 25 Oct |
| CVE-2026-66883 | 6.3 medium | 0.00% | no CPE yet | Improper Handling of Case Sensitivity vulnerability in Erlang Ecosystem Foundation oidcc_plug (Oidcc.Plug.Authorize module) renders the user agent session binding inert, removing a defense in depth control against replay of a stolen session. This vulnerability is associated with program files lib/oidcc/plug/authorize.ex and lib/oidcc/plug/authorization_callback.ex, and program routines Oidcc.Plug.Authorize.call/2 and Oidcc.Plug.AuthorizationCallback.call/2. Oidcc.Plug.Authorize.call/2 reads the initiating client's user agent with get_req_header(conn, "User-Agent"). Plug lowercases incoming header names, but get_req_header/2 matches the supplied key exactly and performs no normalization of its own, so the mixed-case lookup always returns an empty list and nil is written into the session. On the callback side, Oidcc.Plug.AuthorizationCallback treats a stored nil user agent as nothing to compare and returns :ok without inspecting the request. The two behaviours combine so that the check passes unconditionally on every request, including for deployments that explicitly opted in with check_useragent: true, and an authorization callback can be completed from a different user agent than the one that initiated the flow without detection. The check fails open silently, with no error and no log entry, so a deployment cannot tell the binding is absent. The impact is limited to defense in depth. The inert check does not by itself allow an attacker to complete an authorization flow; it removes one layer that would otherwise hinder use of a stolen or leaked session, such as an exfiltrated session cookie replayed from a different client. The CSRF/state, nonce, and PKCE checks are unaffected and continue to function. Deployments that never enabled check_useragent are not affected in practice, since they never expected the binding. The corresponding lookup in Oidcc.Plug.AuthorizationCallback correctly uses the lowercase key and is not affected. This issue affects oidcc_plug: from 0.1.0-alpha.3 before 0.5.0. | 23 Oct |
| CVE-2026-110768 | 6.3 medium | 0.00% | no CPE yet | holaOS through 2026.822.1 contains an authorization bypass vulnerability in the channel-gateway Telegram connector that authorizes senders by mutable username rather than only numeric id. Attackers who register a released username listed in allowFrom can submit turns to the bound workspace agent, which runs configured tools on the operator's machine. | 20 Oct |
| CVE-2025-62788 | 6.3 medium | 0.00% | wazuh | Wazuh is a free and open source platform used for threat prevention, detection, and response. Prior to 4.11.0, w_copy_event_for_log() references memory (initially allocated in OS_CleanMSG()) after it has been freed. A compromised agent can potentially compromise the integrity of the application by sending a specially crafted message to the wazuh manager. An attacker who is able to craft and send an agent message to the wazuh manager can leverage this issue to potentially compromise the integrity of the application (the use of previously freed memory may corrupt valid data, if the memory area in question has been allocated and used properly elsewhere). This vulnerability is fixed in 4.11.0. | 19 Oct |
| CVE-2026-110652 | 6.3 medium | 0.00% | no CPE yet | ONLYOFFICE DocumentServer before 10.0.0 contains an unauthenticated server-side request forgery vulnerability in the /downloadfile/:docid endpoint that allows remote attackers to bypass the private-IP request-filtering-agent by exploiting the isInJwtToken flag when JWT verification is disabled in the default configuration. Attackers can submit a crafted request body containing an internal or loopback URL, which is treated as a DIRECT request skipping private-address filtering, causing the server to fetch and proxy back arbitrary private-network and cloud metadata content. | 16 Oct |
| CVE-2025-64997 | 6.3 medium | 0.00% | checkmk | Insufficient permission validation in Checkmk versions prior to 2.4.0p17 and 2.3.0p42 allow low-privileged users to view agent information via the REST API, which could lead to information disclosure. | 13 Oct |
| CVE-2026-107785 | 6.3 medium | 0.00% | no CPE yet | Crux Agent from 1.9.0 before 2.0.3 uses the full SKA bilocation key as the WireGuard preshared key. When a peering session negotiates use of SHA-512, the key produced is 64 bytes instead of the 32 bytes WireGuard requires. The agent does not validate this size; instead it attempts to use the `wg set` command to update the live tunnel, and write the invalid key to the WireGuard configuration file. The update fails, so the live tunnel keeps using its previous preshared key until the tunnel is shut down. The tunnel will fail to start when restarted. For a peer which has never successfully negotiated a 32-byte bilocation key in a Crux C2 organization which has the "Enforce SKA Use" setting turned off, no preshared key will be set for the tunnel. Therefore, an attacker who is able to intercept and store the peer's traffic, and has access (or will have access) to a cryptographically relevant quantum computer, will be able to decrypt the tunnel. | 9 Oct |
| CVE-2026-88941 | 6.2 medium | 0.00% | kaspersky, network agent, administration server | Kaspersky has fixed a vulnerability in the Network Agent for Windows module of Kaspersky Security Center. The vulnerability could potentially cause the Network Agent module of Kaspersky Security Center to malfunction. There are no evidence of the vulnerability exploitation in-the-wild. | Thu 29 Oct |
| CVE-2026-0248 | 6.2 medium | 0.00% | paloaltonetworks, prisma_access_agent | An improper certificate validation vulnerability in the Prisma Access Agent® for Android and Chrome OS enables an attacker to perform a man-in-the-middle (MitM) attack to intercept VPN traffic. By presenting a certificate for any domain issued by a trusted Certificate Authority, the attacker can capture sensitive device information. The Prisma Access Agent on macOS, Windows, Linux and iOS are not affected. | Mon 26 Oct |
| CVE-2026-57208 | 6.2 medium | 0.00% | no CPE yet | Entire CLI hooks into Git workflows to capture AI agent sessions. Prior to version 0.7.7, a path traversal vulnerability in Entire CLI allows an attacker with push access to the checkpoints repository to craft malicious checkpoint metadata that causes entire session resume or entire checkpoint rewind to write attacker-controlled transcript data outside of the expected session directory. The issue occurs because checkpoint metadata is fetched from the remote entire/checkpoints/v1 branch and the SessionID field was used to construct filesystem paths without validation in the restore path. A malicious SessionID containing absolute paths or path traversal sequences could cause arbitrary files on the victim’s machine to be overwritten. This issue has been patched in version 0.7.7. | 23 Oct |
| CVE-2026-20170 | 6.1 medium | 0.00% | cisco, webex_contact_center | A vulnerability in the Desktop Agent functionality of Cisco Webex Contact Center could have allowed an unauthenticated, remote attacker to conduct cross-site scripting attacks. Cisco has addressed this vulnerability in the Cisco Webex Contact Center service, and no customer action is needed. This vulnerability existed because HTML and script content was not properly handled. Prior to this vulnerability being addressed, an attacker could have exploited this vulnerability by persuading a user to follow a malicious link. A successful exploit could have allowed the attacker to steal sensitive information from the browser, including authentication and session information. | Mon 26 Oct |
| CVE-2026-40112 | 6.1 medium | 0.00% | praison, praisonai | PraisonAI is a multi-agent teams system. Prior to 4.5.128, the Flask API endpoint in src/praisonai/api.py renders agent output as HTML without effective sanitization. The _sanitize_html function relies on the nh3 library, which is not listed as a required or optional dependency in pyproject.toml. When nh3 is absent (the default installation), the sanitizer is a no-op that returns HTML unchanged. An attacker who can influence agent input (via RAG data poisoning, web scraping results, or prompt injection) can inject arbitrary JavaScript that executes in the browser of anyone viewing the API output. This vulnerability is fixed in 4.5.128. | Mon 26 Oct |
| CVE-2026-47250 | 6.1 medium | 0.00% | no CPE yet | mcp-server-kubernetes is a Model Context Protocol server for Kubernetes cluster management. Prior to version 3.7.0, the kubectl_generic tool in mcp-server-kubernetes passes user-supplied flags directly to kubectl without any allowlist, enabling a privilege escalation attack within Kubernetes environments. An attacker who already has limited cluster or codebase access, for example, a developer with pod-deployment permissions but not cluster-admin credentials, can plant a single structured JSON line in an application's log output. When an operator with a privileged kubeconfig uses the MCP server to read those logs and their AI agent follows the injected instruction, kubectl_generic is called with --server=https://attacker.example.com and --insecure-skip-tls-verify=true. kubectl sends all API requests, including the Authorization: Bearer <token> header from the operator's kubeconfig to the attacker's endpoint. The captured token can then be replayed directly against the real Kubernetes API server, granting the attacker the full RBAC permissions of the operator's service account. This issue has been patched in version 3.7.0. | Mon 26 Oct |
| CVE-2026-53765 | 6.1 medium | 0.00% | google, chrome-devtools-mcp | Chrome DevTools for agents (chrome-devtools-mcp) lets your coding agent control and inspect a live Chrome browser. From 0.20.0 until 1.1.0, The chrome-devtools-mcp daemon writes its PID file with fs.writeFileSync() to a deterministic runtime path. On typical macOS environments, and on Linux sessions where $XDG_RUNTIME_DIR is unset, that runtime path falls back to /tmp/chrome-devtools-mcp-<uid>/daemon.pid. Because the write does not use O_NOFOLLOW, a local low-privilege user on the same POSIX host can pre-create /tmp/chrome-devtools-mcp-<victim_uid>/daemon.pid as a symlink to a file writable by the victim. When the victim later starts daemon mode, fs.writeFileSync() follows the symlink and truncates the target file to the daemon PID string. This vulnerability is fixed in 1.1.0. | Sun 25 Oct |
| CVE-2026-53766 | 6.1 medium | 0.00% | google, chrome-devtools-mcp | Chrome DevTools for agents (chrome-devtools-mcp) lets your coding agent control and inspect a live Chrome browser. From 0.24.0 until 1.1.0, McpContext.validatePath() enforces workspace roots by checking whether path.resolve(filePath) textually falls under one of the configured root paths. path.resolve() does not canonicalize symbolic links. As a result, a symlink inside a configured workspace root can point to a file outside that root, pass validation, and then be followed by downstream file read/write operations. This bypass applies even when the MCP client correctly declares the roots capability with a non-empty list. It is separate from the documented legacy behavior where missing roots capability allows all paths. The practical impact is a workspace-boundary bypass. In the write direction, filePath-writing tools can overwrite out-of-root files through an in-root symlink. In the read direction, upload_file can read through the symlink and send the file to the currently selected web page. This vulnerability is fixed in 1.1.0. | Sun 25 Oct |
| CVE-2026-59795 | 6.1 medium | 0.00% | jetbrains, teamcity | In JetBrains TeamCity before 2026.1.2 stored XSS via unauthenticated agent registration was possible | 24 Oct |
| CVE-2026-58413 | 6.1 medium | 0.00% | network-ai | Network-AI is a TypeScript/Node.js multi-agent orchestrator. Prior to version 5.12.2, `EnvironmentManager.restore(env, backupId)` computes the backup path with `join(envDir, '.backups', backupId)` and only checks that this path exists. It does not resolve the result or verify that it remains under `data/<env>/.backups`. A caller can pass a traversal backup ID such as `../../../outside/source-dir` to restore files from an arbitrary directory into the target environment data directory. The issue is fixed in v5.12.2. `restore()` now validates `backupId` against `/^[\w\-]+$/` and asserts `dirname(resolve(join(backupsDir, backupId))) === resolve(backupsDir)` before touching the filesystem. Backup IDs containing path separators or `..` are rejected, so a crafted ID can no longer copy directories from outside `.backups/` into the environment. | 24 Oct |
| CVE-2026-10032 | 6.1 medium | 0.00% | google, a2ui\/web_core | The openUrl function in @a2ui/web_core passes an agent-controlled URL directly to window.open() without validating the URI scheme. A malicious agent can supply a javascript: URI as the url argument of a Button component's functionCall action. When the user clicks the rendered button, arbitrary JavaScript executes in the victim application's browser origin, constituting a stored/reflected XSS with Critical severity. No non-default configuration is required; the Basic Catalog is enabled by default. | 23 Oct |
| CVE-2026-110851 | 6.1 medium | 0.00% | no CPE yet | n8n before 2.41.4 and 2.42.x before 2.42.1 contains an authorization bypass in the Agents create endpoint, which accepts a caller-supplied Agent ID, checks permissions only against the requested project, and updates an existing Agent without verifying which project owns it. An authenticated user with read-only Agent access to another project can supply an existing Agent ID to move that Agent into their own project and rewrite its name, schema and version. | 20 Oct |
| CVE-2026-110841 | 6.1 medium | 0.00% | no CPE yet | n8n before 2.41.4 and 2.42.0 contains an authorization bypass in the agent chat resume endpoint, which accepts a run identifier from the request body without verifying that the suspended checkpoint belongs to the caller. Authenticated project members can list and read other members' private agent conversations through project-scoped endpoints, then approve their pending tool calls, causing the tools to execute and write results into the victim's conversation. | 20 Oct |
| CVE-2025-67794 | 6.1 medium | 0.00% | drivelock | An issue was discovered in DriveLock 24.1 through 24.1.*, 24.2 before 24.2.8, and 25.1 before 25.1.6. Directories and files created by the agent are created with overly permissive ACLs, allowing local users without administrator rights to trigger actions or destabilize the agent. | 13 Oct |
| CVE-2026-91649 | 6.0 medium | 0.00% | oracle, essbase | Vulnerability in Oracle Essbase (component: Java Based Agent). The supported version that is affected is 21.8.3.0.0. Easily exploitable vulnerability allows high privileged attacker with logon to the infrastructure where Oracle Essbase executes to compromise Oracle Essbase. While the vulnerability is in Oracle Essbase, attacks may significantly impact additional products (scope change). Successful attacks of this vulnerability can result in unauthorized access to critical data or complete access to all Oracle Essbase accessible data. CVSS 3.1 Base Score 6.0 (Confidentiality impacts). CVSS Vector: (CVSS:3.1/AV:L/AC:L/PR:H/UI:N/S:C/C:H/I:N/A:N). | Tue 27 Oct |
| CVE-2026-62219 | 6.0 medium | 0.00% | openclaw | OpenClaw 2026.2.12 before 2026.5.26 contain an authorization bypass vulnerability in the hooks allowedAgentIds validation. A lower-trust caller or configured input path can bypass agent ID restrictions by submitting blank agent IDs, allowing actions that should require stronger authorization or policy checks. | 24 Oct |
| CVE-2026-65698 | 6.0 medium | 0.00% | no CPE yet | Void through 1.3.4 contains a path traversal vulnerability in the AI agent file-reading tools that allows network-adjacent attackers to read arbitrary host files outside the open workspace by injecting instructions into content the agent processes. Attackers can supply absolute paths or file:// URIs through the read_file, ls_dir, get_dir_tree, and search_* tools, which lack workspace confinement and bypass the approval gate, enabling silent exfiltration of sensitive files such as SSH private keys or cloud credentials via subsequent tool calls. | 24 Oct |
| CVE-2026-0294 | 6.0 medium | 0.00% | paloaltonetworks, prisma_access_agent | A privilege escalation (PE) vulnerability in the Palo Alto Networks Prisma® Access Agent app on Windows and macOS devices enables a local user to execute code with elevated privileges. The Prisma Access Agent on Linux, iOS, Android, and ChromeOS is not affected. | 23 Oct |
| CVE-2026-108595 | 6.0 medium | 0.00% | no CPE yet | Phi 0.3.0 through 0.28.4 contains a permission bypass vulnerability that allows spawned sub-agents to escape workspace_only_writes and readonly mode by supplying an unchecked workdir to agent_spawn. Attackers can plant prompt-injected instructions in processed content so the agent spawns a worker rooted elsewhere, causing unapproved file writes anywhere the user can write. | 16 Oct |
| CVE-2025-53950 | 6.0 medium | 0.00% | fortinet, fortidlp_agent | An Exposure of Private Personal Information ('Privacy Violation') vulnerability [CWE-359] in Fortinet FortiDLP Agent's Outlookproxy plugin for MacOS and Windows 11.5.1 and 11.4.2 through 11.4.6 and 11.3.2 through 11.3.4 and 11.2.0 through 11.2.3 and 11.1.1. through 11.1.2 and 11.0.1 and 10.5.1 and 10.4.0, and 10.3.1 may allow an authenticated administrator to collect current user's email information. | 9 Oct |
| CVE-2026-115415 | 5.9 medium | 0.00% | no CPE yet | Improper neutralization of input used for LLM prompting in the environment tool in Strands Agents Tools before 0.8.6 might allow remote actors to obtain the agent host's environment variables in cleartext, including cloud credentials and API keys, via a crafted prompt that disables the tool's secret masking. To remediate this issue, users should upgrade to version 0.8.6 or later and rotate any credentials that were present in the environment of agents that loaded the environment tool. | Wed 28 Oct |
| CVE-2026-0246 | 5.9 medium | 0.00% | paloaltonetworks, prisma_access_agent | A vulnerability with a privilege management mechanism in the Palo Alto Networks Prisma Access Agent® enables a locally authenticated non-administrative user to escalate their privileges to root on macOS and Linux or NT AUTHORITY\SYSTEM on Windows. This allows the user to execute arbitrary code and read sensitive information otherwise accessible only to privileged accounts. The Prisma Access Agent on iOS, Android and Chrome OS are not affected. | Mon 26 Oct |
| CVE-2026-0247 | 5.9 medium | 0.00% | paloaltonetworks, prisma_access_agent | Multiple authorization bypass vulnerabilities in the Endpoint DLP component of Prisma Access Agent® allow a local attacker to bypass authentication controls and execute privileged operations. | Mon 26 Oct |
| CVE-2026-54323 | 5.9 medium | 0.00% | no CPE yet | Daytona is a secure and elastic infrastructure runtime for AI-generated code execution and agent workflows. Prior to 0.185.0, the daemon's git clone implementation disabled TLS certificate verification. When a clone request carried Git credentials, the daemon sent the HTTP Basic Authorization header to the remote over a connection whose certificate was never validated, on both the go-git and native git CLI code paths. An attacker able to intercept clone traffic could present any TLS certificate, capture the Git credentials supplied for the clone, and serve tampered repository content into the sandbox. This vulnerability is fixed in 0.185.0. | Sun 25 Oct |
| CVE-2026-58482 | 5.9 medium | 0.00% | network-ai | Network-AI, a TypeScript/Node.js multi-agent orchestrator, has a shipped, exported, documented feature called `ApprovalInbox` (`lib/approval-inbox.ts`). It is the network surface of the human-in-the-loop Approval Gate, which `ApprovalGate` uses to require explicit human approval for high-risk operations. The HTTP server it exposes has no authentication of any kind and sets `Access-Control-Allow-Origin: *` on every route, including the state-changing `POST /approvals/:id/approve` and `/deny`. As a result, in versions 5.0.0 through 5.12.1, any party who can send an HTTP request to the inbox port — a co-located process, a container/SSRF on the same host, a remote client when the operator binds a non-loopback address, or any website the operator visits in a browser (via the wildcard CORS) — can enumerate pending approvals and approve them, defeating the entire human-in-the-loop control and causing the gated high-risk action (e.g. a shell command the agent was holding for review) to execute without consent. This issue is fixed in v5.12.2. `ApprovalInbox` now accepts a `secret` option. When set, the mutating endpoints `POST /:id/approve` and `POST /:id/deny` require an `Authorization: Bearer <secret>` header, validated in constant time with `crypto.timingSafeEqual`. `startServer()` already binds to `127.0.0.1` by default; operators exposing the inbox on a network must set a secret. | 24 Oct |
| CVE-2026-107714 | 5.9 medium | 0.00% | no CPE yet | The Mechanize library is used for automating interaction with websites. Prior to 2.14.1, Mechanize::HTTP::Agent#response_redirect treats redirects as same-origin when the host matches without consistently comparing scheme and port. A same-host HTTPS-to-HTTP redirect can send Authorization and Cookie headers over cleartext, while a same-host redirect to another port can send a caller-supplied Cookie header to a different service. Cookies in Mechanize#cookie_jar remain scoped separately; the issue affects caller-supplied headers and can disclose credentials without affecting integrity or availability. This issue is fixed in version 2.14.1. | 9 Oct |
| CVE-2025-11156 | 5.9 medium | 0.00% | no CPE yet | Netskope was notified about a potential gap in its agent (NS Client) on Windows systems. If this gap is successfully exploited, a local, authenticated user with Administrator privileges can improperly load the driver as a generic kernel service. This triggers the flaw, causing a system crash (Blue-Screen-of-Death) and resulting in a Denial of Service (DoS) for the affected machine. | 8 Oct |
| CVE-2026-0278 | 5.8 medium | 0.00% | paloaltonetworks, prisma_access_agent | Multiple protection mechanism failures in the Prisma Access Agent Data Loss Prevention (DLP) component for Windows allow a local user to bypass DLP policy enforcement controls. The Prisma Access Agent on macOS is not affected. | Sun 25 Oct |
Download these 806 as CSV · subscribe to this list.
EPSS scores: 2026-10-31. NVD cursor: 2026-10-29 00:22. The News page shows only notable CVEs (KEV, your stack, new criticals, EPSS over 50%, or active exploitation per SSVC); this table has all of them.