Vulnerabilities

MLflow versions prior to 3.15.0 contain a server-side request forgery flaw (CWE-918) in the unauthenticated POST /api/2.0/mlflow/webhooks/{id}/test endpoint. The webhook URL is validated only for the original request, but delivery follows redirects and re-resolves the hostname without pinning the validated address, letting an attacker redirect server-side requests to internal network services or cloud instance metadata endpoints. Because the endpoint returns response_status and response_body, attackers can read internal service responses and, per public reporting, steal cloud credentials and secrets from metadata services. Any self-hosted MLflow deployment is affected, with internet-exposed servers at the highest risk. Exploitation is confirmed in the wild: CISA added the flaw to its Known Exploited Vulnerabilities catalog on 2026-08-19, and headlines report active attacks stealing cloud credentials; EPSS puts 30-day exploitation probability at 16.4% (97th percentile).

Do: Upgrade to MLflow 3.15.0 or later immediately, prioritizing any MLflow server reachable from the internet. Until patched, restrict network access to MLflow (especially the /api/2.0/mlflow/webhooks/{id}/test endpoint), place it behind authentication or a reverse proxy, and review webhook logs for unexpected test requests and cloud metadata endpoint access; rotate cloud credentials and secrets if compromise indicators are found. Federal agencies must apply vendor mitigations in accordance with CISA BOD 26-04 guidance or discontinue use if mitigations are unavailable.

The MCP Python SDK, called mcp on PyPI, is a Python implementation of the Model Context Protocol (MCP). From 1.23.0 until 1.27.2, default handlers installed by server.experimental.enable_tasks() for tasks/list, tasks/get, tasks/result, and tasks/cancel operate only on task identifiers without recording the session that created each task, allowing any connected client to enumerate, read results from, consume messages for, or cancel other clients' tasks. This issue is fixed in version 1.27.2.

In MLflow versions prior to 3.14.0, when running with authentication enabled, the trace API endpoints lack proper authorization validators. This allows any authenticated user to bypass experiment-level authorization controls on all trace operations, including reading, deleting, and modifying traces on experiments they do not have permission to access. The issue arises from the `_before_request` handler, which does not register authorization validators for trace endpoints, resulting in requests proceeding without validation. This vulnerability can expose sensitive data, destroy audit logs, and allow unauthorized modifications.

A vulnerability has been found in MLflow up to 4666cffc7912ea606d592fc38d6a75e2935f65e7. The impacted element is an unknown function of the component Experiment-scoped Label Schema CRUD API. Such manipulation leads to missing authorization. It is possible to launch the attack remotely. A high complexity level is associated with this attack. The exploitability is regarded as difficult. The exploit has been disclosed to the public and may be used. A reply to the GitHub issue explains, that "[t]he labeling schema PR has not been merged yet. The auth handlers will be added before the release."

A flaw has been found in MLflow up to 3.10.0. This issue affects the function mlflow.data.digest_utils of the file mlflow/data/digest_utils.py of the component Dataset Digest Computation. This manipulation causes use of weak hash. It is possible to launch the attack on the local host. The attack is considered to have high complexity. The exploitability is assessed as difficult. The exploit has been published and may be used. The project was informed of the problem early through a pull request but has not reacted yet.

A vulnerability in mlflow/mlflow versions prior to 3.11.0 allows for the resolution of environment variables in AI Gateway secrets, which can be exploited to exfiltrate sensitive server-side environment credentials to an attacker-controlled endpoint. This issue arises because the `api_key` field in gateway secrets can accept `$ENV_VAR` references, which are resolved against the MLflow server's environment during runtime. The resolved secrets are then sent in provider authentication headers to the configured upstream `api_base`. This vulnerability can be exploited by low-privileged authenticated users in basic-auth deployments or by unauthenticated users in default deployments without `basic-auth`. The impact includes potential leakage of sensitive credentials such as cloud artifact credentials (`AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY`), which could lead to artifact poisoning and cross-boundary code execution in downstream environments. The issue is fixed in version 3.11.0.

MLflow 3.9.0 with basic-auth (`--app-name basic-auth`) fails to enforce authorization checks for multiple Gateway API 'list' endpoints. Specifically, the `BEFORE_REQUEST_HANDLERS` dictionary in `mlflow/server/auth/__init__.py` does not include entries for `ListGatewaySecretInfos`, `ListGatewayEndpoints`, and `ListGatewayModelDefinitions`. This allows any authenticated user, regardless of their assigned permissions, to enumerate all gateway secrets, endpoints, and model definitions. This vulnerability exposes sensitive information, such as API keys, endpoint configurations, and proprietary model definitions, to unauthorized users.

A vulnerability in MLflow versions <=3.10.1.dev0 allows unauthorized access to multipart upload (MPU) endpoints when the `--serve-artifacts` mode is enabled. The authorization logic does not enforce resource-level permission checks for `/mlflow-artifacts/mpu/*` endpoints, enabling attackers to overwrite artifacts belonging to other users. This can lead to unauthorized cross-user writes, model supply chain poisoning, and arbitrary code execution when compromised models are loaded. The issue is resolved in version 3.10.0.

In mlflow/mlflow versions up to 3.9.0, the `SearchModelVersions` REST API endpoint and the `mlflowSearchModelVersions` GraphQL query lack proper per-model authorization checks when basic authentication is enabled. This allows any authenticated user to enumerate all model versions across all registered models, regardless of their permission level. The issue arises due to the absence of `SearchModelVersions` in the `BEFORE_REQUEST_VALIDATORS` and `AFTER_REQUEST_HANDLERS` for the REST API, and its omission from `GraphQLAuthorizationMiddleware.PROTECTED_FIELDS` for GraphQL. This vulnerability can expose sensitive information such as model names, version descriptions, source URIs, tags, and other metadata, potentially revealing proprietary or confidential details in multi-tenant environments. The issue is resolved in version 3.10.0.

In MLflow version 3.9.0, the MLflow Assistant feature introduced improper origin validation in its /ajax-api endpoints. This vulnerability allows a remote attacker to exploit cross-origin requests from a malicious webpage to interact with the MLflow Assistant running on a victim's local machine. By bypassing the loopback-only restriction, the attacker can modify the Assistant's configuration to enable full access, which in turn allows the execution of arbitrary commands via the Claude Code sub-agent. This issue is resolved in version 3.10.0.

In mlflow/mlflow versions prior to 3.11.0, the `get_or_create_nfs_tmp_dir()` function in `mlflow/utils/file_utils.py` creates temporary directories with world-writable permissions (0o777), and the `_create_model_downloading_tmp_dir()` function in `mlflow/pyfunc/__init__.py` creates directories with group-writable permissions (0o770). These insecure permissions allow local attackers to tamper with model artifacts, such as cloudpickle-serialized Python objects, and achieve arbitrary code execution when the tampered artifacts are deserialized via `cloudpickle.load()`. This vulnerability is particularly critical in environments with shared NFS mounts, such as Databricks, where NFS is enabled by default. The issue is a continuation of the vulnerability class addressed in CVE-2025-10279, which was only partially fixed.

MLflow 3.9.0 and earlier, when launched with authentication enabled (--app-name basic-auth) and served via uvicorn (ASGI), fails to enforce authentication on most FastAPI routes because the permission middleware (_find_fastapi_validator()) only covers /gateway/ paths (CWE-305). An unauthenticated remote attacker can therefore submit jobs, read job results, and cancel running jobs through the Job API (/ajax-api/3.0/jobs/*), and can inject arbitrary OpenTelemetry trace data into experiments via /v1/traces. Only deployments that run MLflow with basic-auth enabled over an ASGI (uvicorn) server are affected; servers using the default Flask serving path or running without authentication enabled are not impacted by this bypass. The flaw is fixed in MLflow 3.10.0. No confirmed in-the-wild exploitation is known at this time, but a public proof-of-concept exists and EPSS assigns a ~21% probability of exploitation within 30 days (97th percentile).

Do: Upgrade to MLflow 3.10.0 or later. In the interim, identify any tracking servers launched with --app-name basic-auth behind uvicorn (ASGI) and restrict network access to /ajax-api/3.0/jobs/* and /v1/traces — for example via an authenticating reverse proxy or firewall rules. Also review job history and experiment traces for signs of unauthenticated job submissions or injected traces.

The MCP Registry provides MCP clients with a list of MCP servers, like an app store for MCP servers. Prior to 1.7.7, the Registry's HTTP-based namespace verification (POST /v0/auth/http, POST /v0.1/auth/http) uses safeDialContext (internal/api/handlers/v0/auth/http.go:67-110) to refuse dialling private/internal addresses when fetching the well-known public-key file from a publisher-supplied domain. The blocklist (isBlockedIP, lines 125-133) relies entirely on Go stdlib's IsLoopback / IsPrivate / IsLinkLocalUnicast / IsMulticast / IsUnspecified plus a manual CGNAT range. None of these cover IPv6 6to4 (2002::/16), NAT64 (64:ff9b::/96 and 64:ff9b:1::/48 per RFC 8215), or deprecated site-local (fec0::/10) — all of which encode arbitrary IPv4 in the address bits and tunnel to RFC1918 / cloud-metadata services on dual-stack / NAT64-enabled hosts. This vulnerability is fixed in 1.7.7.

A vulnerability in the `_create_model_version()` handler of `mlflow/server/handlers.py` in mlflow/mlflow versions 3.9.0 and earlier allows an unauthenticated remote attacker to read arbitrary files from the server's filesystem. The issue arises when a `CreateModelVersion` request includes the tag `mlflow.prompt.is_prompt`, which bypasses source path validation. This enables an attacker to store an arbitrary local filesystem path as the model version source. The `get_model_version_artifact_handler()` function later uses this source to serve files without verifying the model version's prompt status, leading to a complete confidentiality compromise. This issue is fixed in version 3.10.0.

Zarf is an Airgap Native Packager Manager for Kubernetes. Versions 0.23.0 through 0.74.1 contain an arbitrary file write vulnerability in the zarf package inspect sbom and zarf package inspect documentation subcommands. These subcommands output file paths are constructed by joining a user-controlled output directory with the package's Metadata.Name field read directly from the untrusted package's zarf.yaml manifest. Although Metadata.Name is validated against a regex on package creation, an attacker can unarchive a package to modify the Metadata.Name field to contain path traversal sequences such as ../../etc/cron.d/malicious or absolute paths like /home/user/.ssh/authorized_keys, along with the corresponding files inside SBOMS.tar. This allows writing attacker-controlled content to arbitrary filesystem locations within the permissions of the user running the inspect command. This issue has been fixed in version 0.74.2.

MCP Java SDK is the official Java SDK for Model Context Protocol servers and clients. Prior to 1.0.0, the java-sdk contains a DNS rebinding vulnerability. This vulnerability allows an attacker to access a locally or network-private java-sdk MCP server via a victims browser that is either local, or network adjacent. This allows an attacker to make any tool call to the server as if they were a locally running MCP connected AI agent. This vulnerability is fixed in 1.0.0.

MLflow is vulnerable to an authorization bypass affecting the AJAX endpoint used to download saved model artifacts. Due to missing access‑control validation, a user without permissions to a given experiment can directly query this endpoint and retrieve model artifacts they are not authorized to access. This issue affects MLflow version through 3.10.1

In mlflow/mlflow, the FastAPI job endpoints under `/ajax-api/3.0/jobs/*` are not protected by authentication or authorization when the `basic-auth` app is enabled. This vulnerability affects the latest version of the repository. If job execution is enabled (`MLFLOW_SERVER_ENABLE_JOB_EXECUTION=true`) and any job function is allowlisted, any network client can submit, read, search, and cancel jobs without credentials, bypassing basic-auth entirely. This can lead to unauthenticated remote code execution if allowed jobs perform privileged actions such as shell execution or filesystem changes. Even if jobs are deemed safe, this still constitutes an authentication bypass, potentially resulting in job spam, denial of service (DoS), or data exposure in job results.

The Go MCP SDK used Go's standard encoding/json. Prior to version 1.4.0, the Model Context Protocol (MCP) Go SDK does not enable DNS rebinding protection by default for HTTP-based servers. When an HTTP-based MCP server is run on localhost without authentication with StreamableHTTPHandler or SSEHandler, a malicious website could exploit DNS rebinding to bypass same-origin policy restrictions and send requests to the local MCP server. This could allow an attacker to invoke tools or access resources exposed by the MCP server on behalf of the user in those limited circumstances. This issue has been patched in version 1.4.0.

MCP Java SDK is the official Java SDK for Model Context Protocol servers and clients. Prior to versions 0.83.0, 1.0.1, and 1.1.1, there is a hardcoded wildcard CORS vulnerability. This issue has been patched in versions 0.83.0, 1.0.1, and 1.1.1.

A command injection vulnerability exists in mlflow/mlflow when serving a model with `enable_mlserver=True`. The `model_uri` is embedded directly into a shell command executed via `bash -c` without proper sanitization. If the `model_uri` contains shell metacharacters, such as `$()` or backticks, it allows for command substitution and execution of attacker-controlled commands. This vulnerability affects the latest version of mlflow/mlflow and can lead to privilege escalation if a higher-privileged service serves models from a directory writable by lower-privileged users.

A command injection vulnerability exists in MLflow's model serving container initialization code, specifically in the `_install_model_dependencies_to_env()` function. When deploying a model with `env_manager=LOCAL`, MLflow reads dependency specifications from the model artifact's `python_env.yaml` file and directly interpolates them into a shell command without sanitization. This allows an attacker to supply a malicious model artifact and achieve arbitrary command execution on systems that deploy the model. The vulnerability affects versions 3.8.0 and is fixed in version 3.8.2.

MCP Ruby SDK is the official Ruby SDK for Model Context Protocol servers and clients. Prior to version 0.9.2, the Ruby SDK's streamable_http_transport.rb implementation contains a session hijacking vulnerability. An attacker who obtains a valid session ID can completely hijack the victim's Server-Sent Events (SSE) stream and intercept all real-time data. Version 0.9.2 contains a patch.

In the latest version of mlflow/mlflow, when the `basic-auth` app is enabled, tracing and assessment endpoints are not protected by permission validators. This allows any authenticated user, including those with `NO_PERMISSIONS` on the experiment, to read trace information and create assessments for traces they should not have access to. This vulnerability impacts confidentiality by exposing trace metadata and integrity by allowing unauthorized creation of assessments. Deployments using `mlflow server --app-name=basic-auth` are affected.

The Go MCP SDK used Go's standard encoding/json. Prior to version 1.4.1, the Go SDK's Streamable HTTP transport accepted browser-generated cross-site `POST` requests without validating the `Origin` header and without requiring `Content-Type: application/json`. In deployments without Authorization, especially stateless or sessionless configurations, this allows an arbitrary website to send MCP requests to a local server and potentially trigger tool execution. Version 1.4.1 contains a patch for the issue.

A vulnerability in MLflow's pyfunc extraction process allows for arbitrary file writes due to improper handling of tar archive entries. Specifically, the use of `tarfile.extractall` without path validation enables crafted tar.gz files containing `..` or absolute paths to escape the intended extraction directory. This issue affects the latest version of MLflow and poses a high/critical risk in scenarios involving multi-tenant environments or ingestion of untrusted artifacts, as it can lead to arbitrary file overwrites and potential remote code execution.

A command injection vulnerability exists in mlflow/mlflow versions before v3.7.0, specifically in the `mlflow/sagemaker/__init__.py` file at lines 161-167. The vulnerability arises from the direct interpolation of user-supplied container image names into shell commands without proper sanitization, which are then executed using `os.system()`. This allows attackers to execute arbitrary commands by supplying malicious input through the `--container` parameter of the CLI. The issue affects environments where MLflow is used, including development setups, CI/CD pipelines, and cloud deployments.

Agentgateway is an open source data plane for agentic AI connectivity within or across any agent framework or environment. Prior to version 0.12.0, when converting MCP tools/call request to OpenAPI request, input path, query, and header values are not sanitized. This issue has been patched in version 0.12.0.

Zarf is an Airgap Native Packager Manager for Kubernetes. From version 0.54.0 to before version 0.73.1, a path traversal vulnerability in archive extraction allows a specifically crafted Zarf package to create symlinks pointing outside the destination directory, enabling arbitrary file read or write on the system processing the package. This issue has been patched in version 0.73.1.

The Go MCP SDK used Go's standard encoding/json.Unmarshal for JSON-RPC and MCP protocol message parsing in versions prior to 1.3.1. Go's standard library performs case-insensitive matching of JSON keys to struct field tags — a field tagged json:"method" would also match "Method", "METHOD", etc. This violated the JSON-RPC 2.0 specification, which defines exact field names. A malicious MCP peer may have been able to send protocol messages with non-standard field casing that the SDK would silently accept. This had the potential for bypassing intermediary inspection and coss-implementation inconsistency. Go's standard JSON unmarshaling was replaced with a case-sensitive decoder in commit 7b8d81c. Users are advised to update to v1.3.1 to resolve this issue.

Model Context Protocol Servers is a collection of reference implementations for the model context protocol (MCP). In mcp-server-git versions prior to 2026.1.14, the git_add tool did not validate that file paths provided in the files argument were within the repository boundaries. Because the tool used GitPython's repo.index.add() rather than the Git CLI, relative paths containing `../` sequences that resolve outside the repository were accepted and staged into the Git index. Users are advised to upgrade to 2026.1.14 or newer to remediate this issue.

Valkey-Bloom is a Rust based Valkey module which brings a Bloom Filter (Module) data type into the Valkey distributed key-value database. Prior to commit a68614b6e3845777d383b3a513cedcc08b3b7ccd, a specially crafted `RESTORE` command can cause Valkey to hit an assertion, causes the server to shutdown. Valkey modules are required to handle errors in RDB parsing by using `VALKEYMODULE_OPTIONS_HANDLE_IO_ERRORS` flag. If this flag is not set, errors encountered during parsing result in a system assertion which shuts down the system. Even though the Valkey-bloom module correctly handled the parsing, it did not originally set the flag. Commit a68614b6e3845777d383b3a513cedcc08b3b7ccd contains a patch. One may mitigate this defect by disabling the `RESTORE` command if it is unused by one's application.

Valkey is a distributed key-value database. Prior to versions 9.0.2, 8.1.6, 8.0.7, and 7.2.12, a malicious actor with access to the Valkey clusterbus port can send an invalid packet that may cause an out bound read, which might result in the system crashing. The Valkey clusterbus packet processing code does not validate that a clusterbus ping extension packet is located within buffer of the clusterbus packet before attempting to read it. Versions 9.0.2, 8.1.6, 8.0.7, and 7.2.12 fix the issue. As an additional mitigation, don't expose the cluster bus connection directly to end users, and protect the connection with its own network ACLs.

MCP TypeScript SDK is the official TypeScript SDK for Model Context Protocol servers and clients. From version 1.10.0 to 1.25.3, cross-client response data leak when a single McpServer/Server and transport instance is reused across multiple client connections, most commonly in stateless StreamableHTTPServerTransport deployments. This issue has been patched in version 1.26.0.

In mlflow version 2.20.3, the temporary directory used for creating Python virtual environments is assigned insecure world-writable permissions (0o777). This vulnerability allows an attacker with write access to the `/tmp` directory to exploit a race condition and overwrite `.py` files in the virtual environment, leading to arbitrary code execution. The issue is resolved in version 3.4.0.

MLFlow versions up to and including 3.4.0 are vulnerable to DNS rebinding attacks due to a lack of Origin header validation in the MLFlow REST server. This vulnerability allows malicious websites to bypass Same-Origin Policy protections and execute unauthorized calls against REST endpoints. An attacker can query, update, and delete experiments via the affected endpoints, leading to potential data exfiltration, destruction, or manipulation. The issue is resolved in version 3.5.0.

Anthropic's MCP TypeScript SDK versions up to and including 1.25.1 contain a regular expression denial of service (ReDoS) vulnerability in the UriTemplate class when processing RFC 6570 exploded array patterns. The dynamically generated regular expression used during URI matching contains nested quantifiers that can trigger catastrophic backtracking on specially crafted inputs, resulting in excessive CPU consumption. An attacker can exploit this by supplying a malicious URI that causes the Node.js process to become unresponsive, leading to a denial of service.

Model Context Protocol Servers is a collection of reference implementations for the model context protocol (MCP). In mcp-server-git versions prior to 2025.9.25, the git_init tool accepted arbitrary filesystem paths and created Git repositories without validating the target location. Unlike other tools which required an existing repository, git_init could operate on any directory accessible to the server process, making those directories eligible for subsequent git operations. The tool was removed entirely, as the server is intended to operate on existing repositories only. Users are advised to upgrade to 2025.9.25 or newer to remediate this issue.

The MCP Python SDK, called `mcp` on PyPI, is a Python implementation of the Model Context Protocol (MCP). Prior to version 1.23.0, tThe Model Context Protocol (MCP) Python SDK does not enable DNS rebinding protection by default for HTTP-based servers. When an HTTP-based MCP server is run on localhost without authentication using FastMCP with streamable HTTP or SSE transport, and has not configured TransportSecuritySettings, a malicious website could exploit DNS rebinding to bypass same-origin policy restrictions and send requests to the local MCP server. This could allow an attacker to invoke tools or access resources exposed by the MCP server on behalf of the user in those limited circumstances. Note that running HTTP-based MCP servers locally without authentication is not recommended per MCP security best practices. This issue does not affect servers using stdio transport. This vulnerability is fixed in 1.23.0.