fsspec: Server-Side Template Injection in ReferenceFileSystem leads to Remote Code Execution

Summaryfsspec.implementations.reference.ReferenceFileSystem parses a "references"JSON document Kerchunk format supplied either inline or via a URL. Theparser renders fields from this JSON through un-sandboxedjinja2.Template....render... calls in three locations. An attacker whocontrols the JSON...

python-jose algorithm confusion guard bypassed by DER-encoded public keys

python-jose through 3.5.0 fails to properly validate asymmetric keys in HMAC initialization, accepting DER-encoded public keys that lack PEM armor or SSH prefixes. Attackers holding the service's public key can forge HS256 tokens that pass verification when algorithms are not explicitly restricte...

Headroom vulnerable to Cross-Site WebSocket Hijacking (CSWSH)

SummaryThe Headroom WebSocket server does not validate the Origin header of incoming client WebSocket requests before forwarding the request to the upstream server, allowing malicious WebSocket clients to perform arbitrary LLM requests without authentication. This can be exploited by a malicious...

GPT Researcher MCP STDIO configuration allows remote command execution

An issue in Open Source GPT Researcher v3.3.7 allows attackers to execute arbitrary commands on a victim system via user interaction with a crafted HTML page...

geopy: Regular Expression Denial of Service (ReDoS) in geopy.Point

Impactgeopy.Point and Point.fromstring may take excessive CPU time when parsing long, malformed coordinate strings due to inefficient regular-expression behavior. The numeric Point constructor is not affected.Geocoders' reverse methods called with string inputs exercise the vulnerable...

BlenderMCP path traversal in download_polyhaven_asset allows arbitrary file writes

BlenderMCP before commit 30a3308 contains a path traversal vulnerability in the downloadpolyhavenasset method that allows attackers to write arbitrary files by injecting traversal sequences in API response include keys. Attackers performing MITM attacks or prompt injection can supply malicious...

DeepTutor missing MCP tool authorization allows non-admin users to invoke unrestricted tools

DeepTutor before version 1.4.10 contains an authorization bypass vulnerability that allows low-privilege users to invoke unrestricted MCP tools due to the allowedmcptools function returning None instead of a denied result when mcptools is omitted from a user's grant in...

LlamaIndex DuckDBVectorStore vulnerable to SQL injection through ref_doc_id

An SQL injection vulnerability exists in the delete function of DuckDBVectorStore in run-llama/llamaindex version v0.12.19. This vulnerability allows an attacker to manipulate the refdocid parameter, enabling them to read and write arbitrary files on the server, potentially leading to remote code...

Scrapy: Arbitrary Module Import via Referrer-Policy Header in RefererMiddleware

ImpactSince version 1.4.0, Scrapy respects the Referrer-Policy response header to decide whether and how to set a Referer header on follow-up requests.If the header value looked like a valid Python import path, Scrapy would import the referenced object and call it, assuming it referred to a...

PYSEC-2026-4319

A security flaw has been discovered in vllm-project vLLM up to 0.31.0. This impacts the function gettokenbincountsandmask of the file vllm/modelexecutor/layers/utils.py of the component Penalty Handler. Performing a manipulation results in denial of service. Remote exploitation of the attack is...

PYSEC-2026-4199

vLLM is an inference and serving engine for large language models. Prior to 0.30.0, structured-output request failures can escape request-scoped validation and reach the EngineCore fatal-error path. A per-request backend mismatch can re-raise a grammar compilation exception, padding produced by t...

PYSEC-2026-4325

vLLM is an inference and serving engine for large language models. Prior to 0.30.0, the Rust frontend's trackhttpmetrics middleware records the raw HTTP method token as a Prometheus label for requests reaching registered routes. An unauthenticated attacker can send unique arbitrary method tokens ...

PYSEC-2026-4198

vLLM is an inference and serving engine for large language models. Prior to 0.30.0, flash late-interaction scoring at the /score and /rerank endpoints derives each worker's querykey value from the caller-controlled X-Request-Id header. A concurrent request that reuses a victim's identifier can...

PYSEC-2026-4197

vLLM is an inference and serving engine for large language models. Prior to 0.28.0, the default mirrored multimodal LRU cache can commit a media hash in the frontend sender cache during multimodal rendering and before engine admission, while the engine receiver cache never receives the payload if...

PYSEC-2026-4196

Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.82.0 until 2.118.1, HTMLBackendOptionsrenderpage=True permits file URLs because HTMLDocumentBackend.getbrowserrequestblockreason does not enforce the...

PYSEC-2026-4191

Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.94.0 until 2.132.0, callers that opt into LatexBackendOptionstikzengine="tectonic" invoke docling/backend/latex/engines/tectonic.py to compile an untrusted TikZ bo...

PYSEC-2026-4193

Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.83.0 until 2.131.0, the KServeV2OcrModel class defined in docling/models/stages/ocr/kservev2ocrmodel.py sends page images to its configured endpoint without checki...

PYSEC-2026-4195

Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.0.0 until 2.131.0, the HTML, JATS, OpenDocument spreadsheet, and BoxNote backends, including docling/backend/htmlbackend.py, docling/backend/jatsbackend.py, and...

PYSEC-2026-4190

Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.91.0 until 2.132.0, validateurlsafety in docling/backend/utils/imageresourceloader.py validates a hostname with a single IPv4 lookup and then allows the HTTP clien...

PYSEC-2026-4194

Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.45.0 until 2.131.0, METS-GBS format detection in docling/datamodel/document.py and the backend in docling/backend/metsgbsbackend.py call tarfile.TarFile.getmembers...

PYSEC-2026-4192

Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.27.0 until 2.131.0, Docling plugin factories in docling/models/factories/basefactory.py call loadsetuptoolsentrypoints before applying the allowexternalplugins...

PYSEC-2026-4189

Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. From 2.95.0 until 2.132.0, the HTML image resource loader in docling/backend/utils/imageresourceloader.py forwards headers configured through the HTMLBackendOptions.heade...

JupyterLab: Cross-site scripting (XSS) in JupyterLab via notebook cells pasted from the system clipboard

DescriptionJupyterLab 4.5.0 enabled copying and pasing cells through the system clipboard. The paste path parses clipboard text as cell JSON and inserts the cells without clearing metadata.trusted, so a payload can declare its own output as trusted. JupyterLab does not sanitize a trusted output a...

JupyterLab: Cross-site scripting (XSS) in JupyterLab via notebook cells pasted from the system clipboard

DescriptionJupyterLab 4.5.0 enabled copying and pasing cells through the system clipboard. The paste path parses clipboard text as cell JSON and inserts the cells without clearing metadata.trusted, so a payload can declare its own output as trusted. JupyterLab does not sanitize a trusted output a...

JupyterLab: Argument injection in JupyterLab extension uninstall exposes server-readable files and internal URLs

JupyterLab's PyPI extension manager runs python -m pip uninstall with the extension name taken from the request body. ExtensionHandler.post validates the name for cmd=install but not for cmd=uninstall, so a name that begins with - reaches the command line and pip reads it as an option rather than...

JupyterLab: Cross-site scripting (XSS) in JupyterLab via notebook cells pasted from the system clipboard

DescriptionJupyterLab 4.5.0 enabled copying and pasing cells through the system clipboard. The paste path parses clipboard text as cell JSON and inserts the cells without clearing metadata.trusted, so a payload can declare its own output as trusted. JupyterLab does not sanitize a trusted output a...

JupyterLab: Cross-site scripting (XSS) in JupyterLab via crafted language package (jupyterlab.json)

DescriptionA language pack ships a Plural-Forms header saying how the language counts, for example nplurals=2; plural=n != 1;. JupyterLab turns that string into a function with new Function, so the header gets executed. The check that meant to keep it safe was a regular expression. The regex was...

JupyterLab: Cross-site scripting (XSS) in JupyterLab via crafted language package (jupyterlab.json)

DescriptionA language pack ships a Plural-Forms header saying how the language counts, for example nplurals=2; plural=n != 1;. JupyterLab turns that string into a function with new Function, so the header gets executed. The check that meant to keep it safe was a regular expression. The regex was...

JupyterLab: Cross-site scripting (XSS) in JupyterLab via notebook cells pasted from the system clipboard

DescriptionJupyterLab 4.5.0 enabled copying and pasing cells through the system clipboard. The paste path parses clipboard text as cell JSON and inserts the cells without clearing metadata.trusted, so a payload can declare its own output as trusted. JupyterLab does not sanitize a trusted output a...

pypdf: Possible long runtimes with large amount of embedded files

ImpactAn attacker who uses this vulnerability can craft a PDF which leads to long runtimes. This requires accessing the embedded files through the dictionary-based API. PatchesThis has been fixed in pypdf==6.19.0. WorkaroundsIf you cannot upgrade yet, consider applying the changes from PR 4081...

pypdf: Possible long runtimes for partially malformed FlateDecode streams (Follow-up)

ImpactAn attacker who uses this vulnerability can craft a PDF which leads to long runtimes. This requires a partially malformed /FlateDecode stream, where the byte-by-byte decompression is used. PatchesThis has been fixed in pypdf==6.18.1. WorkaroundsIf you cannot upgrade yet, consider applying t...

pypdf: Possible long runtimes when generating appearance streams

ImpactAn attacker who uses this vulnerability can craft a PDF which leads to long runtimes. This requires updating form field values with flattening enabled, which triggers appearance stream generation. PatchesThis has been fixed in pypdf==6.19.0. WorkaroundsIf you cannot upgrade yet, consider...

pypdf: Possible large memory usage when retrieving alphabetical page labels

ImpactAn attacker who uses this vulnerability can craft a PDF which leads to large memory consumption. This requires accessing the page labels of a document with large alphabetical labels. PatchesThis has been fixed in pypdf==6.19.0. WorkaroundsIf you cannot upgrade yet, consider applying the...

urllib3: Chunked Deflate streaming can enter an infinite loop

Impacturllib3's streaming API is designed for the efficient handling of large HTTP responses by reading content in chunks instead of loading the entire response body into memory at once.urllib3 can decompress response bodies according to the HTTP Content-Encoding header. When streaming a...

pypdf: Possible large memory usage for large /ToUnicode streams (Follow-up 2)

ImpactAn attacker who uses this vulnerability can craft a PDF which leads to large memory consumption. This requires parsing the /ToUnicode entry of a font with unusually large values, for example during text extraction. PatchesThis has been fixed in pypdf==6.18.1. WorkaroundsIf you cannot upgrad...

urllib3: HTTPResponse.stream()/read_chunked() buffers an unbounded chunk-size line into memory

Impacturllib3's streaming API is designed for efficiently handling large HTTP responses by reading the content in chunks, rather than loading the entire response body into memory at once. When decoding a chunked-transfer-encoded response, this API reads each chunk's size field by buffering until ...

pypdf: Possible large memory usage when retrieving Roman page labels

ImpactAn attacker who uses this vulnerability can craft a PDF which leads to large memory consumption. This requires accessing the page labels of a document with large Roman numerals. PatchesThis has been fixed in pypdf==6.17.0. WorkaroundsIf you cannot upgrade yet, consider applying the changes...

pypdf: Possible long runtimes/large memory usage when parsing indirect objects

ImpactAn attacker who uses this vulnerability can craft a PDF which leads to long runtimes. This requires reading a PDF document with long indirect object headers, not terminated by whitespace. PatchesThis has been fixed in pypdf==6.18.0. WorkaroundsIf you cannot upgrade yet, consider applying th...

pypdf: Possible large memory usage when parsing font data

ImpactAn attacker who uses this vulnerability can craft a PDF which leads to large memory consumption. This requires parsing the /Widths entry of a TrueType or Type1 fonts with unusually large values, for example during text extraction. PatchesThis has been fixed in pypdf==6.18.1. WorkaroundsIf y...

LiteLLM: Authenticated SSRF and provider-credential exfiltration via unvalidated request-body routing parameters

ImpactAny authenticated LiteLLM proxy user could redirect an outbound provider call to a destination they control and cause the proxy to send its own configured provider credentials to that destination. The proxy's request-body validation was a denylist that did not cover every sensitive paramete...

PyJWT.decode() reintroduces options-dict mutation, enabling silent claim-verification bypass on dict reuse

SummaryPyJWT.decode/decodecomplete mutates a caller-supplied options dict in place whenever verifysignature is falsy, adding verifyexp/verifynbf/verifyiat/verifyaud/verifyiss/verifysub/verifyjti keys directly onto that object. If application code reuses the same options dict across calls a config...

PyJWT: Unauthenticated RecursionError DoS in pre-verification payload parse (PyJWKClient.get_signing_key_from_jwt / verify_signature=False)

SummaryPyJWKClient.getsigningkeyfromjwttoken — the first step of the JWKSverification flow documented in docs/usage.rst — must decode a token'spayload before its signature can be checked, viajwt.apijwt.decodecompletetoken, options="verifysignature": False.That call parses the payload with...

PyJWT: PyJWK accepts empty HMAC keys, bypassing PyJWT's empty-key validation

SummaryA service that verifies HS256 tokens using an empty oct JWK through PyJWK, including a PyJWK obtained from PyJWKSet, can therefore accept attacker-generated tokens as authenticated.PyJWT 2.13.0 rejects an empty HMAC key when it is supplied through the raw str/bytes key path, but accepts th...

PyJWT: Asymmetric-PEM detection bypass: whitespace/line-ending-mutated public keys skip the HS/asymmetric confusion guard

Prerequisites both conditions must hold; both are deployment properties, not attacker-controlled at request time:- The jwt.decode allow-list mixes an HMAC algorithm with an asymmetric one, e.g. algorithms="ES256", "HS256" the RFC 8725 footgun the guard exists to backstop.- The verification key is...

PyJWT: Malformed RSA JWK aborts parsing of an entire JWK Set

SummaryA malformed RSA JWK inside a JWK Set aborts parsing of the entire set instead of being skipped, because RSAAlgorithm.fromjwk can raise a plain ValueError that isn't caught by PyJWKSet's per-key error-skipping logic. Affected component / version- Package: PyJWT PyPI, ecosystem pip- Files:...

PyJWT: Uncaught RecursionError in jwt.decode() on deeply nested token header

Packagepyjwt PyPI Affected versionstested & verified on: 2.13.0. Every version whose PyJWS.load translates only ValueError is affected DescriptionPyJWS.looad jwt/apijws.py:337 splits the compact token, base64url decodes the header segment and hands it to json.loads before verifysignature runs. Th...

PyJWT: PyJWKClient still amplifies unauthenticated JWKS fetches on unknown kid values (incomplete fix of CVE-2026-48524)

SummaryCVE-2026-48524 GHSA-fhv5-28vv-h8m8, "PyJWKClient unbounded JWKS endpoint requests via attacker-controlled kid values DoS" was fixed in 2.13.0 by stopping fetchdata from clearing the cache on a fetch error. That closed one amplification path but did not add the mitigation the advisory's tit...

PyJWT accepts public JWK containers as HMAC secrets

SummaryPyJWT 2.13.0 contains an incomplete defense against algorithm confusion whenan application mixes symmetric and asymmetric algorithms in one verificationpath. A public RSA, EC, or OKP JWK can be accepted as an HMAC secret when itis wrapped in a JWKS object, nested in an array, or represente...

PyJWT: ReDoS vulnerability when calling the `is_pem_format` function.

SummaryThere is a Re-DoS vulnerability in the ispemformat function which results in a intesive CPU usage if an attacker is been able to provide a custom certificate. DetailsThe problem is that the lazy quantifier .+? will always first try to match as little as possible until it finds a ---- END...

urllib3: HTTPS proxy TLS configuration may be ignored or overridden

Impacturllib3 supports configuring TLS independently for an HTTPS proxy and the target server.proxysslcontext, proxyasserthostname, and proxyassertfingerprint configure the TLS connection to the proxy. sslcontext and the other target-specific TLS parameters configure the connection to the target...