fast-uri versions 2.3.1 through 3.1.2 and 4.0.0 fail to canonicalize Unicode (IDN) hostnames for HTTP-family URLs. The IDN conversion path calls a helper that does not exist on the global URL constructor, silently leaving the host in its original Unicode form while normalize() and equal() still return values that differ from a WHATWG-compatible URL parser. Applications that use fast-uri to enforce host-based policy (denylists, loopback filtering, redirect validation, outbound proxy routing) before passing the same URL to Node's URL or fetch can be bypassed when the two implementations resolve the same input to different hosts. Patches: upgrade to fast-uri 3.1.3 for the 3.x line or 4.0.1 for the 4.x line. Workarounds: enforce host policy using the same URL parser used for the actual request, or reject non-ASCII hosts before policy checks.

jackson-databind contains the general-purpose data-binding functionality and tree-model for Jackson Data Processor. From 2.10.0 until 2.18.8, 2.21.4, and 3.1.4, BasicPolymorphicTypeValidator.Builder.allowIfSubTypeIsArray() allowlists any array type based only on clazz.isArray(), without validating the array's component (element) type against the configured allowlist. A PTV built with allowIfSubTypeIsArray() plus an explicit concrete-type allowlist therefore still permits EvilType[] even though EvilType is not allowlisted. When Jackson deserializes the elements and no per-element type IDs are present, it instantiates the component type directly with no further PTV check, bypassing the allowlist. This vulnerability is fixed in 2.18.8, 2.21.4, and 3.1.4.

vLLM is an inference and serving engine for large language models (LLMs). From 0.3.0 until 0.22.0, a vulnerability in ASGI web servers and starlette's trust on those web servers enables an authentication bypass of the OpenAI API AuthenticationMiddleware. It allows to use the API without providing the configured VLLM_API_KEY or --api-key. This vulnerability is fixed in 0.22.0.

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. Prior to 1.84.0, a Host-header parsing flaw in the LiteLLM proxy could, under specific conditions, allow unauthenticated access to protected management routes. The auth layer derived the effective route from request.url.path in litellm/proxy/auth/auth_utils.py::get_request_route(), which Starlette reconstructs from the Host header. A crafted Host could therefore make the auth gate evaluate a different route from the one FastAPI dispatched. This vulnerability is fixed in 1.84.0.

NLTK (Natural Language Toolkit) is a suite of open source Python modules, data sets, and tutorials supporting research and development in Natural Language Processing. Prior to 3.10.0-rc1, nltk.data.load() in NLTK is vulnerable to path traversal via URL-encoded path separators and traversal segments when using the nltk: URL scheme. The unsafe-path regex check is performed before url2pathname() decodes the %xx sequences (a classic decode-after-check / TOCTOU-style flaw), allowing an attacker to bypass the protection documented in NLTK's SECURITY.md and read arbitrary files from the filesystem. While literal traversal strings such as ../../../etc/passwd are correctly blocked, encoded variants such as %2fetc%2fpasswd, %2e%2e%2f..., and ..%2f..%2f slip past the regex and are subsequently decoded into a real filesystem path. This vulnerability is fixed in 3.10.0-rc1.

A weakness has been identified in BerriAI litellm up to 1.59.8. Affected is the function UserAPIKeyAuth of the file litellm/proxy/_experimental/mcp_server/auth/user_api_key_auth_mcp.py of the component MCP Proxy. Executing a manipulation can lead to improper authentication. The attack may be launched remotely. The exploit has been made available to the public and could be used for attacks. The vendor was contacted early about this disclosure.

A flaw was found in the AWX GitHub webhook integration. When processing GitHub pull_request webhooks, the controller stores the pull_request.statuses_url value from the webhook payload without validating that it points to a trusted GitHub API endpoint. If a job template is configured with a GitHub Personal Access Token as its webhook credential, the controller later POSTs that token to the stored callback URL when posting job status updates. An attacker who can submit a correctly signed forged webhook using the job template's webhook_key can redirect the callback to an attacker-controlled URL and exfiltrate the configured GitHub PAT.

Starlette is a lightweight ASGI framework/toolkit. In versions 1.0.1 and earlier, StaticFiles on Windows is vulnerable to SSRF. An UNC path such as \\attacker.com\share can cause os.path.realpath to initiate an outbound SMB connection before the path is rejected, exposing the service account’s NTLMv2 credentials for offline cracking or relay even though the HTTP response is only a 404. The issue affects default follow_symlink=False deployments, including frameworks built on Starlette such as FastAPI; POSIX systems and follow_symlink=True are unaffected. The issue is fixed in 1.1.0.

ws is an open source WebSocket client and server for Node.js. All versions from 1.1.0 up to (but not including) 5.2.5, from 6.0.0 up to 6.2.4, from 7.0.0 up to 7.5.11, and from 8.0.0 up to 8.21.0 are affected by a memory exhaustion DoS vulnerability. A peer can send a high volume of exceptionally small fragments and data chunks, with modest network traffic, to force the remote peer into allocating and holding structural wrappers that consume far more memory than the default documented message-size limit, leading to process termination due to OOM. This issue has been fixed in versions 5.2.5, 6.2.4, 7.5.11, and 8.21.0.

A command injection vulnerability was found in galaxy_ng. The do_git_checkout() function in the legacy role import API (v1) interpolates unsanitized git ref names (branch/tag names) into shell commands executed via subprocess.run() with shell=True. An authenticated user who controls a git repository can create a branch or tag with shell metacharacters in the name to achieve remote code execution on the pulp worker. The vulnerable endpoint is only reachable when GALAXY_ENABLE_LEGACY_ROLES is set to True, which is not the default configuration.

A flaw was found in Ansible Lightspeed. This vulnerability, related to insufficient session expiration, allows a remote attacker to maintain persistent access to the Ansible Lightspeed instance. If an attacker exfiltrates a valid OAuth (Open Authorization) access token before a user logs out, they can continue to authenticate and access sensitive data. This is because the application fails to invalidate the token on the backend, leaving it valid until its natural expiration. This can lead to unauthorized read access to Ansible resources such as inventories, playbooks, and configuration data.

form-data is a library for creating readable multipart/form-data streams. In versions through 4.0.5, the `field` argument to `FormData#append` and the `filename` option are concatenated verbatim into the `Content-Disposition` header without escaping carriage return (CR), line feed (LF), or double-quote (") characters. An application that passes attacker-controlled data as a field name or filename (for example, an API gateway that turns JSON object keys into multipart field names) allows the attacker to terminate the header line and inject additional headers, or to smuggle entire additional multipart parts, into the request the application forwards to a backend. This can let the attacker add or override form fields (e.g. set `is_admin=true`) seen by the downstream parser. This is an instance of CWE-93 (CRLF injection). The fix escapes CR, LF, and `"` as `%0D`, `%0A`, and `%22` in field names and filenames, matching the serialization browsers use per the WHATWG HTML multipart/form-data encoding algorithm. Exploitation requires the consuming application to use untrusted input as a field name or filename; applications that use only fixed/trusted field names are not affected. Fixed in 2.5.6, 3.0.5, and 4.0.6.

Axios is a promise based HTTP client for the browser and Node.js. Prior to 0.32.0 and 1.16.0, Axios’ Node.js HTTP adapter can leak proxy credentials to a redirect target in affected versions. When a request is sent through an authenticated proxy, Axios may add a Proxy-Authorization header. If Axios then follows a redirect and the redirected request is no longer sent through that proxy, the stale Proxy-Authorization header can remain on the redirected request and be sent to the redirect target. This affects Node.js's use of Axios with automatic redirects enabled and an authenticated proxy configuration. Browser adapters are not affected. This vulnerability is fixed in 0.32.0 and 1.16.0.

Axios is a promise based HTTP client for the browser and Node.js. Prior to 0.32.0 and 1.16.0, Axios’s Node.js HTTP adapter may forward a Proxy-Authorization header to a redirected origin during specific proxy-to-direct redirect flows. This affects Node.js usage, where an initial HTTP request is sent through an authenticated HTTP proxy, redirects are followed, and the redirected URL is no longer proxied. Under affected redirect shapes, the final origin can receive the proxy credential that was intended only for the outbound proxy. This vulnerability is fixed in 0.32.0 and 1.16.0.

Axios is a promise based HTTP client for the browser and Node.js. Axios versions 1.7.0 through 1.15.x did not enforce configured request and response size limits when requests were sent with the fetch adapter. Applications that selected adapter: 'fetch', or ran in environments where axios resolved to the fetch adapter, could receive or send bodies larger than maxContentLength or maxBodyLength despite those limits being explicitly configured. This can cause resource exhaustion in server-side usage when a malicious or compromised server returns an oversized response, when an attacker can supply a large data: URL, or when an application forwards attacker-controlled request bodies through axios while relying on maxBodyLength as a boundary. This vulnerability is fixed in 0.32.0 and 1.16.0.

Axios is a promise based HTTP client for the browser and Node.js. Axios versions before 0.32.0 on the 0.x line and before 1.16.0 on the 1.x line build a regular expression from the configured XSRF cookie name without escaping regex metacharacters. In standard browser environments, an attacker who can influence the cookie name passed to axios can cause expensive regex backtracking while axios reads document.cookie. The practical impact is client-side availability degradation, such as freezing the affected browser tab while axios prepares a request. The issue does not affect ordinary Node.js HTTP adapter usage, React Native, or web workers, where axios does not read document.cookie. This vulnerability is fixed in 0.32.0 and 1.16.0.

Axios is a promise based HTTP client for the browser and Node.js. From 0.19.0 to before 0.31.1 and 1.15.2, Axios contains prototype-pollution gadgets in request config processing. If another vulnerability in the same JavaScript process has already polluted Object.prototype.transformResponse, affected Axios versions may treat that inherited value as request configuration or as an option validator. Axios does not itself create the prototype pollution. Exploitability requires a separate prototype-pollution vulnerability or equivalent attacker control over Object.prototype before Axios creates a request. This vulnerability is fixed in 0.31.1 and 1.15.2.

Axios is a promise based HTTP client for the browser and Node.js. From 1.0.0 to before 1.16.0, the Axios library is vulnerable to a Prototype Pollution "Gadget" attack that allows any Object.prototype pollution in the application's dependency tree to be escalated into a full Man-in-the-Middle (MITM) attack — intercepting, reading, and modifying all HTTP traffic including authentication credentials. The HTTP adapter at lib/adapters/http.js:670 reads config.proxy via standard property access, which traverses the prototype chain. Because proxy is not present in Axios defaults, the merged config object has no own proxy property, making it trivially injectable via prototype pollution. Once injected, setProxy() routes all HTTP requests through the attacker's proxy server. This vulnerability is fixed in 1.16.0.

Axios is a promise based HTTP client for the browser and Node.js. Prior to 0.32.0 and 1.16.0, Axios does not normalise IPv4-mapped IPv6 addresses. When NO_PROXY lists an IPv4 address such as 127.0.0.1 or 169.254.169.254, a request URL using the IPv4-mapped IPv6 form (::ffff:7f00:1, ::ffff:a9fe:a9fe) still routes through the configured proxy. Node.js resolves these addresses to the underlying IPv4 host, so the request reaches the internal service via the proxy rather than being blocked. This vulnerability is fixed in 0.32.0 and 1.16.0.

JavaScript Cookie is a JavaScript API for handling cookies, client-side. Prior to version 3.0.7, js-cookie's internal assign() helper copies properties with for...in + plain assignment. When the source object is produced by JSON.parse, the JSON object's "__proto__" member is an own enumerable property, so the for…in enumerates it and the target[key] = source[key] write triggers the Object.prototype.__proto__ setter on the fresh target ({}). The result is a per-instance prototype hijack: Object.prototype itself is untouched, but the merged attributes object now inherits attacker-controlled keys. Because the consuming set() function then enumerates the merged object with another for...in, every key the attacker placed on the polluted prototype lands in the resulting Set-Cookie string as an attribute pair. The attacker can set domain=, secure=, samesite=, expires=, and path= on cookies whose attributes the developer thought were locked down. This issue has been patched in version 3.0.7.

A path traversal vulnerability was found in awxkit, the CLI tool for AWX. The YAML !include directive does not sanitize file paths, allowing an attacker to craft a malicious YAML file that reads arbitrary YAML-formatted files from the local filesystem when a user imports it using "awx --conf.format yaml import". This is a client-side vulnerability requiring user interaction.

A flaw was found in ansible-core. The ansible-galaxy role install command processes dependency specifications from a role's meta/requirements.yml file. Due to improper neutralization of argument delimiters, a malicious role author can inject arbitrary git configuration flags through the src field. This allows arbitrary code execution on the machine of a user who installs the role via ansible-galaxy role install.

A vulnerability in the LightGlue model loading path of huggingface/transformers version 5.2.0 allows an attacker-controlled model repository to execute arbitrary code during model initialization. The issue arises because the `trust_remote_code` parameter, intended to prevent remote code execution, is overridden by untrusted serialized configuration data in a nested code path. Specifically, when loading a LightGlue model using `AutoModel.from_pretrained()` with `trust_remote_code=False`, the `LightGlueConfig` reads the `trust_remote_code` value from the untrusted `config.json` file and propagates it into nested `AutoConfig.from_pretrained()` calls. This results in the execution of attacker-provided Python modules, even when the victim explicitly disables remote code execution. The vulnerability poses a high risk for environments such as API inference servers, research notebooks, CI/CD pipelines, and model evaluation workers, potentially leading to credential theft, lateral movement, or persistence/backdoor deployment.

(*x509.Certificate).VerifyHostname previously called matchHostnames in a loop over all DNS Subject Alternative Name (SAN) entries. This caused strings.Split(host, ".") to execute repeatedly on the same input hostname. With a large DNS SAN list, verification costs scaled quadratically based on the number of SAN entries multiplied by the hostname's label count. Because x509.Verify validates hostnames before building the certificate chain, this overhead occurred even for untrusted certificates.

AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to version 3.14.0, using ``CookieJar.load()`` with untrusted input may allow arbitrary code execution. Most applications using this function will be doing so with the user's own data, so this is unlikely to affect many applications. Version 3.14.0 patches the issue. If an application does allow attacker controlled files to be loaded, a workaround on older releases would be to sanitize the files before loading.

launch-editor allows users to open files with line numbers in editor from Node.js. Prior to version 2.9.0, due to the insufficient sanitization of the `file` argument in the `launchEditor`, an attacker can execute arbitrary commands on Windows by supplying a filename that contains special characters. This issue has been fixed in the `launch-editor` version 2.9.0, corresponding to vite version 5.4.9.

pip would treat console_scripts and gui_scripts as paths instead of file names without sanitizing the resolved absolute path to the installation directory, leading to entry points being installed outside the installation directory.

opentelemetry-java is the Java implementation of the OpenTelemetry API for recording telemetry, and SDK for managing telemetry recorded by the API. Prior to 1.62.0, a vulnerability affects the baggage propagation implementation in opentelemetry-api and opentelemetry-extension-trace-propagators. Parsing oversized baggage causes unbounded memory allocation and CPU consumption. Because baggage is automatically re-injected into every outgoing request, the effect can fan out to downstream services that never received the original malicious request. This vulnerability is fixed in 1.62.0.

PyJWT is a JSON Web Token implementation in Python. Prior to 2.13.0, when the verifier is decoding JSON Web Tokens, while supporting both asymmetric and HMAC algorithms, the library does not validate use of JSON Web Keys in HMAC algorithm, allowing attacker to use the issuer public key as the secret key for HMAC algorithm. This vulnerability is fixed in 2.13.0.

Starlette is a lightweight ASGI framework/toolkit. Prior to version 1.0.1, the HTTP `Host` request header was not validated before being used to reconstruct `request.url`. Because the routing algorithm relies on the raw HTTP path while `request.url` is rebuilt from the `Host` header, a malformed header could make `request.url.path` differ from the path that was actually requested. Middleware and endpoints that apply security restrictions based on `request.url` (rather than the raw `scope` path) could therefore be bypassed. Users should upgrade to a version greater than or equal to version 1.0.1, which validates the `Host` header against the grammar of RFC 9112 §3.2 / RFC 3986 §3.2.2 when constructing `request.url` and falls back to `scope["server"]` for malformed values.

Banks generates meaningful LLM prompts using a template language that makes sense. Prior to 2.4.2, banks uses jinja2.Environment() (unsandboxed) to render prompt templates. Applications that pass user-supplied strings as the template argument to Prompt() are vulnerable to Server-Side Template Injection (SSTI), which can lead to Remote Code Execution (RCE) on the host system. This vulnerability is fixed in 2.4.2.

The ToASCII and ToUnicode functions incorrectly accept Punycode-encoded labels that decode to an ASCII-only label. For example, ToUnicode("xn--example-.com") incorrectly returns the name "example.com" rather than an error. This behavior can lead to privilege escalation in programs using the idna package. For example, a program which performs privilege checks on the ASCII hostname may reject "example.com" but permit "xn--example-.com". If that program subsequently converts the ASCII hostname to Unicode, it will inadvertently permits access to the Unicode name "example.com".

shell-quote's `quote()` function did not validate object-token inputs against the operator model used by `parse()`. The `.op` field was backslash-escaped character by character using `/(.)/g`, which in JavaScript does not match line terminators (\n, \r, U+2028, U+2029). A line terminator in `.op` therefore passed through unescaped into the output; POSIX shells treat a literal newline as a command separator, so any content after it would execute as a second command. The vulnerable code path is reachable in two ways: (1) direct construction of `{ op: '...\n...' }` from external input, and (2) via `parse(cmd, envFn)` when `envFn` returns object tokens whose `.op` is attacker-influenced. Both are documented API surface. Fixed by replacing the per-character escape with strict shape validation: `.op` must match the parser's control-operator allowlist; `{ op: 'glob', pattern }` validates `pattern` and forbids line terminators; `{ comment }` validates `comment` and forbids line terminators; any other object shape throws `TypeError`.

LiteLLM prior to 1.83.10 allows a user to modify their own user_role via the /user/update endpoint. While the endpoint correctly restricts users to updating only their own account, it does not restrict which fields may be changed. A user who can reach this endpoint can set their role to proxy_admin, gaining full administrative access to LiteLLM including all users, teams, keys, models, and prompt history. Users with the org_admin role have legitimate access to this endpoint and can exploit this vulnerability without chaining any additional flaw.

LiteLLM prior to 1.83.14 allows an authenticated internal_user to create API keys with access to routes that their role does not permit. When generating a key, the allowed_routes field is stored without verifying that the specified routes fall within the user's own permissions. A key created with access to admin-only routes can then be used to reach those routes successfully, bypassing the role-based access controls that would otherwise block the request, enabling full privilege escalation from internal_user to proxy_admin.

The JSONSerializer and CBORSerializer in APScheduler (all versions including 3.10.x and 4.0.0a5) are vulnerable to Remote Code Execution (RCE) via Insecure Deserialization. The unmarshal_object function allows for arbitrary class instantiation and state injection by dynamically importing modules and calling __setstate__ on any class available in the Python environment. An attacker can exploit this by submitting a specially crafted JSON or CBOR payload to an application using these serializers

python jsonpickle 2.0.0 contains a remote code execution vulnerability that allows attackers to execute arbitrary Python commands by deserializing malicious JSON payloads containing py/repr objects. Attackers can craft JSON strings with py/repr directives that invoke the eval function during deserialization to execute arbitrary code.

ws is an open source WebSocket client and server for Node.js. Prior to 8.20.1, the websocket.close() implementation is vulnerable to uninitialized memory disclosure when a TypedArray is passed as the reason argument. This vulnerability is fixed in 8.20.1.

Python-Multipart is a streaming multipart parser for Python. Prior to 0.0.27, python-multipart has a denial of service vulnerability in multipart part header parsing. When parsing multipart/form-data, MultipartParser previously had no limit on the number of part headers or the size of an individual part header. An attacker could send a request with either many repeated headers without terminating the header block or a single very large header value, causing excessive CPU work before request rejection or completion. This vulnerability is fixed in 0.0.27.

urllib3 is an HTTP client library for Python. From 2.6.0 to before 2.7.0, urllib3 could decompress the whole response instead of the requested portion (1) during the second HTTPResponse.read(amt=N) call when the response was decompressed using the official Brotli library or (2) when HTTPResponse.drain_conn() was called after the response had been read and decompressed partially (compression algorithm did not matter here). These issues could cause urllib3 to fully decode a small amount of highly compressed data in a single operation. This could result in excessive resource consumption (high CPU usage and massive memory allocation for the decompressed data) on the client side. This vulnerability is fixed in 2.7.0.

protobufjs compiles protobuf definitions into JavaScript (JS) functions. Prior to 7.5.6 and 8.0.2, protobufjs could recurse without a depth limit while decoding nested protobuf data. This affected both skipping unknown group fields and generated decoding of nested message fields. A crafted protobuf binary payload could cause the JavaScript call stack to be exhausted during decoding. This vulnerability is fixed in 7.5.6 and 8.0.2.

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.81.16 to before version 1.83.7, a database query used during proxy API key checks mixed the caller-supplied key value into the query text instead of passing it as a separate parameter. An unauthenticated attacker could send a specially crafted Authorization header to any LLM API route (for example POST /chat/completions) and reach this query through the proxy's error-handling path. An attacker could read data from the proxy's database and may be able to modify it, leading to unauthorised access to the proxy and the credentials it manages. This issue has been patched in version 1.83.7.

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.80.5 to before version 1.83.7, the POST /prompts/test endpoint accepted user-supplied prompt templates and rendered them without sandboxing. A crafted template could run arbitrary code inside the LiteLLM Proxy process. The endpoint only checks that the caller presents a valid proxy API key, so any authenticated user could reach it. Depending on how the proxy is deployed, this could expose secrets in the process environment (such as provider API keys or database credentials) and allow commands to be run on the host. This issue has been patched in version 1.83.7.

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.74.2 to before version 1.83.7, two endpoints used to preview an MCP server before saving it — POST /mcp-rest/test/connection and POST /mcp-rest/test/tools/list — accepted a full server configuration in the request body, including the command, args, and env fields used by the stdio transport. When called with a stdio configuration, the endpoints attempted to connect, which spawned the supplied command as a subprocess on the proxy host with the privileges of the proxy process. The endpoints were gated only by a valid proxy API key, with no role check. Any authenticated user — including holders of low-privilege internal-user keys — could therefore run arbitrary commands on the host. This issue has been patched in version 1.83.7.

Axios is a promise based HTTP client for the browser and Node.js. From version 1.0.0 to before version 1.15.2, fFive config properties (auth, baseURL, socketPath, beforeRedirect, and insecureHTTPParser) in the HTTP adapter are read via direct property access without hasOwnProperty guards, making them exploitable as prototype pollution gadgets. When Object.prototype is polluted by another dependency in the same process, axios silently picks up these polluted values on every outbound HTTP request. This issue has been patched in version 1.15.2.

Well-crafted inputs reaching ParseAddress, ParseAddressList, and ParseDate were able to trigger excessive CPU exhaustion and memory allocations.

When using LookupCNAME with the cgo DNS resolver, a very long CNAME response can trigger a double-free of C memory and a crash.

Pathological inputs could cause DoS through consumePhrase when parsing an email address according to RFC 5322.

fast-uri normalize() decoded percent-encoded authority delimiters inside the host component and then re-emitted them as raw delimiters during serialization. A host that combined an allowed domain, an encoded at-sign, and a different domain was re-emitted with the at-sign as a raw userinfo separator, changing the URI's authority to the second domain. Applications that normalize untrusted URLs before host allowlist checks, redirect validation, or outbound request routing can be steered to a different authority than the input appeared to specify. Versions <= 3.1.1 are affected. Update to 3.1.2 or later.

fast-uri decoded percent-encoded path separators and dot segments before applying dot-segment removal in its normalize() and equal() functions. Encoded path data was treated like real slashes and parent-directory references, so distinct URIs could collapse onto the same normalized path. Applications that normalize or compare attacker-controlled URLs to enforce path-based policy can be bypassed, with a path that appears confined under an allowed prefix normalizing to a different location. Versions <= 3.1.0 are affected. Update to 3.1.1 or later.