Vulnerabilities (CVE)

n8n versions before 1.123.73, 2.35.4, and 2.36.2 contain a remote code execution vulnerability in the Git node. The node reset a fixed list of command-bearing configuration keys before each operation, but that list did not cover the content-filter and merge-driver key families. A repository with local configuration setting one of those keys together with a matching attribute pattern causes git to execute the configured command during an ordinary Add, Commit, Checkout, or Pull operation. The command runs as the n8n process user.Show less

n8n before 2.35.4 and 2.36.x before 2.36.2 contain a query injection vulnerability in the Elasticsearch Document Get All and Google Cloud Firestore Document Query operations, which build their JSON query by interpolating expression values directly into the query string before parsing. A value containing quote and brace characters can close the intended field and introduce new query operators, turning an intended single-document lookup into a full-collection read.Show less

n8n before 2.35.4 and 2.36.x before 2.36.2 does not validate credential references in the inline workflow JSON of nodes that execute an inline sub-workflow (e.g., the Workflow Tool node). A shared-workflow editor, or any user creating/updating a workflow via the REST API, Public API, or MCP, can persist a node referencing a credential they do not own. When the workflow is later executed under an identity that holds the credential, the inline sub-workflow resolves the secret and can send it to an attacker-controlled endpoint, resulting in credential exfiltration.Show less

n8n versions before 2.36.2 contain an expression sandbox bypass vulnerability where free identifiers in spread, computed-key, switch-case, or class-extension positions resolve against process globals. Authenticated users with workflow-edit permission can mutate host objects through expression evaluation, with changes persisting process-wide until restart.Show less

Kimai (kimai/kimai) before 2.65.0 contains a business logic / improper authorization vulnerability in the default team creation endpoints. An authenticated user with project permission-management privileges can create or use a customer, project, or activity whose name matches an existing team; because the endpoints POST /api/customers/{id}/team, POST /api/projects/{id}/team, and POST /api/activities/{id}/team reuse an existing team of the same name and add the current user as teamlead without verifying that the user is authorized to manage that team, the attacker gains unauthorized team-lead (administration) rights over the existing team. Fixed in 2.65.0.Show less

Nodemailer before 8.0.4 is vulnerable to SMTP command injection through the unsanitized envelope.size parameter. When an application passes a custom envelope object with a size property containing CRLF characters to sendMail(), the value is concatenated into the SMTP MAIL FROM command (as SIZE=...) without sanitization, allowing injection of arbitrary SMTP commands such as RCPT TO to silently add attacker-controlled recipients. Exploitation requires the application to expose the envelope size to attacker-controlled input, as Nodemailer does not include size in the default auto-constructed envelope.Show less

CodeWhale versions >= 0.3.27 and < 0.8.64 (published as deepseek-tui >= 0.3.27 and < 0.8.41 on npm and 0.3.27 through 0.8.41 on crates.io, and as codewhale / codewhale-tui >= 0.8.41 and < 0.8.64) contain an argument injection vulnerability in the git_show tool. The model-supplied rev parameter is passed unvalidated into the git show argv without an --end-of-options sentinel, so a value beginning with --output= is interpreted as a git flag. Because the tool is registered as auto-approved and advertised as read-only, an attacker (via a malicious repository combined with prompt injection) can cause an unprompted arbitrary file write at the privilege of the invoking user, targeting sensitive files such as ~/.ssh/authorized_keys, ~/.bashrc, or ~/.gitconfig. Fixed in codewhale / codewhale-tui 0.8.64 by adding rev validation; the deprecated deepseek-tui crate has no fixed release.Show less

CodeWhale versions >= 0.3.10 and < 0.8.64 (published as deepseek-tui >= 0.3.10 and < 0.8.41 on npm and 0.3.10 through 0.8.41 on crates.io, and as codewhale / codewhale-tui >= 0.8.41 and < 0.8.64) contain a vulnerability in the exec_shell_interact (alias exec_interact) tool, whose approval_requirement returns ApprovalRequirement::Auto. This overrides the default Required approval for code-executing tools, so LLM-controlled stdin is written into an already-approved long-running interactive shell (e.g., a python3 -i REPL, mysql, ssh, or sudo -i session) without any approval prompt. An attacker who can inject instructions via untrusted content the agent ingests (a fetched page, MCP result, or repo file) can cause commands to run at the privilege level of that approved process. Fixed in codewhale / codewhale-tui 0.8.64; the deprecated deepseek-tui crate has no fixed release.Show less

SiYuan's development branch (endpoint introduced by commit 9b8e8956f, not present in v3.7.3 or master, patched in v3.7.4-alpha.3) contains a missing-authorization vulnerability in the /api/av/getAttributeViewSearchTarget endpoint. The route is registered with CheckAuth only and performs no authorization checks (no CheckReadonly, no publish-access or encrypted-notebook gating). Given a database identifier taken from a published page and a keyword, an anonymous reader can query the endpoint to retrieve matching database row content, including rows that publish filters (FilterAttributeViewByPublishAccess) would otherwise withhold. No released stable version is affected.Show less

actix-http versions before 3.12.1 contain an HTTP request smuggling vulnerability in the HTTP/1.1 parser that accepts requests with both Content-Length and Transfer-Encoding: chunked headers. Unauthenticated remote attackers can exploit this through a front-end intermediary to desynchronize backend requests and smuggle malicious HTTP requests to the Actix service.Show less

The Flex Objects plugin (through 1.4.6, tested with Grav 2.0.11) contains an incorrect authorization vulnerability in its Flex Objects API. FlexApiController::update() checks only the general Flex directory permission and does not apply the additional target/field/super-admin checks enforced by the dedicated Users and Groups API controllers. An authenticated account with api.access, admin.login, and users.update permissions (but without api.users.write or admin.super) can use the generic /api/v1/flex-objects/user-accounts endpoint to change a super administrator's password, or the /api/v1/flex-objects/user-groups endpoint to grant its group admin.super, resulting in full site takeover. Fixed in Flex Objects 1.4.7.Show less

The getgrav/grav-plugin-api plugin before 1.0.13 fails to validate that the scopes of a newly created API key are a subset of the caller's scopes in createApiKey. The self-target path of requireApiKeyPermission() requires only the baseline api.access scope, and the new key's scopes are read directly from the request body with no subset check. An attacker holding a minimal-scope API key on a super account can submit an empty scopes array to mint an unscoped, full-access super key, bypassing scope restrictions (and enabling further chains such as configuration write to RCE).Show less

The getgrav/grav-plugin-api plugin before 1.0.13 contains an API-key scope cap bypass in the POST /reports/twig-content/allowlist endpoint (ReportsController). The endpoint enforces requirePermission('api.config.write') followed by a bare isSuperAdmin() check instead of requireSuper(). Because isSuperAdmin() reads access.api.super directly and never consults api_key_scopes, a least-privilege API key scoped to api.config.write minted on a super account passes the gate, allowing an attacker to append attacker-chosen tokens to the security.twig_sandbox allowlist (persisted to user/config/security.yaml). Widening the allowlist turns any subsequent Twig-in-content render into an SSTI/RCE sink.Show less

The actix-files crate (actix_files) before version 0.6.10 contains an information exposure vulnerability. When a non-existing folder is passed as the serve_from argument to Files::new(), the mount path defaults to an empty path; the service then joins the request path with this empty path and canonicalizes it, causing Rust to resolve it as a relative path. As a result, an attacker can request paths that resolve relative to the application's working directory and access unintended files.Show less

actix-files before 0.6.10 contains a denial of service vulnerability triggered by an empty Range header in GET requests for static files. When panic is set to abort, remote attackers can crash the process on-demand by sending a GET request with an empty Range header.Show less

SiYuan v3.7.4-alpha.1 (a pre-release; the endpoint does not exist in stable v3.7.3 or earlier) contains an information disclosure vulnerability in the /api/av/getAttributeViewFieldViews endpoint. The route is registered with CheckAuth only and applies no publish-access filtering, so reader-role callers can retrieve the complete database view structure — every view's name, icon, layout type, and per-field Hidden flag — for any database whose avID is supplied, regardless of authorization. The issue was introduced by commit acfc02ee8 and fixed in v3.7.4-alpha.2.Show less

n8n before 1.123.67, 2.31.5, and 2.32.1 contains a SQL injection vulnerability in the PostgresTrigger node, which interpolates user-supplied identifier parameters (channel, function, and trigger names) into SQL statements without proper escaping. An authenticated user can inject arbitrary SQL executed against the connected PostgreSQL database with the configured credential's privileges, allowing full read and write access.Show less

n8n before 1.123.67, 2.31.5, and 2.32.1 contains a credential authorization bypass in the HTTP Request node. An authenticated member with edit access to a shared workflow can reference another user's credential while specifying the credential type via an expression. Because the pre-execution permission check compares the unresolved expression instead of the resolved credential type, the ownership check is skipped and the credential is loaded at execution time, allowing the member to use or exfiltrate a credential they were not granted. Exploitation requires knowing the target credential's identifier.Show less

n8n before 2.31.5 and 2.32.x before 2.32.1 contain a path-confinement bypass in the @n8n/computer-use file-search (search_files) tool. A crafted search pattern can bypass the base-directory confinement check and expand to locations outside the configured directory, causing the tool to return the names and contents of arbitrary local files readable by the daemon's OS user. Any deployment where an actor can influence the tool's search input is affected.Show less

n8n before 2.32.1 (and before 2.31.5) is vulnerable to account takeover via the Token Exchange Embed Login feature. When a validly-signed incoming token was matched to a local account by its email claim, the service did not verify that the email claim was verified, nor that the trusted key's permitted role ceiling covered that account. As a result, anyone able to obtain a token accepted by a configured trusted key (for example, a trusted issuer emitting unverified email addresses) could authenticate as any existing user and gain full account control. This issue only affects instances where the embed login feature is enabled and at least one trusted key source is configured.Show less

n8n versions before 2.32.1 fail to enforce the Allowed HTTP Request Domains allowlist in multiple AI and LLM nodes when user-supplied base or endpoint URLs are configured. Low-privileged workflow editors with use-only access to shared credentials can redirect requests to attacker-controlled hosts and exfiltrate credential secrets for reuse against underlying services.Show less

llama.cpp builds b1886 through b7445 contain a race condition use-after-free vulnerability in the LLaMA-Android JNI wrapper where bench_1model() and free_1context() lack synchronization, allowing Thread A to operate on freed memory while Thread B concurrently frees the llama_context. Attackers can exploit this by performing heap spray with attacker-controlled data containing a fake vtable to hijack the vtable pointer at offset +0x30, causing llama_batch_allocr::clear() to dereference arbitrary memory and achieve remote code execution.Show less

llama.cpp builds b1886 through b7445 contain a null pointer dereference vulnerability in the LLaMA-Android JNI wrapper where the bench_1model() function fails to validate the model context pointer before dereferencing it. Attackers can supply a malicious, corrupt, or truncated model file to trigger a null context condition, causing a SIGSEGV crash that terminates the Android application process and results in denial of service.Show less

llama.cpp builds b1886 through b7445 contain an integer overflow vulnerability in the LLaMA-Android JNI wrapper where the new_1batch() function multiplies sizeof(llama_seq_id) by an attacker-controlled n_seq_max parameter without overflow validation, causing heap buffer allocation to wrap and allocate insufficient memory. Attackers can exploit this by providing a crafted n_seq_max value through a malicious model file or JNI call to trigger heap corruption and achieve denial of service or arbitrary code execution on Android applications using the LLaMA-Android binding.Show less

radare2-mcp version 1.6.0 and earlier contains an os command injection vulnerability that allows remote attackers to execute arbitrary commands by bypassing the command filter through shell metacharacters in user-controlled input passed to r2_cmd_str(). Attackers can inject shell metacharacters through the jsonrpc interface parameters to achieve remote code execution on the host running radare2-mcp without requiring authentication.Show less