Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of Requests, when making requests through a Requests `Session`, if the first request is made with `verify=False` to disable cert verification, all subsequent requests to the same host will continue to ignore cert verification regardless of changes to the value of `verify`. This behavior will continue for the lifecycle of the connection in the connection pool. Requests 2.32.0 fixes the issue, but versions 2.32.0 and 2.32.1 were yanked due to conflicts with CVE-2024-35195 mitigation.

Affected versions of the requests package are vulnerable to Insecure Temporary File reuse due to predictable temporary filename generation in extract_zipped_paths(). The requests.utils.extract_zipped_paths() utility extracts files from zip archives into the system temporary directory using a deterministic path, and if that file already exists, the function reuses it without validating that it is the expected extracted content.

Affected versions of Requests are vulnerable to proxy credential leakage. When redirected to an HTTPS endpoint, the Proxy-Authorization header is forwarded to the destination server due to the use of rebuild_proxies to reattach the header. This may allow a malicious actor to exfiltrate sensitive information.

Requests is an HTTP library. Due to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. Users should upgrade to version 2.32.4 to receive a fix. For older versions of Requests, use of the .netrc file can be disabled with `trust_env=False` on one's Requests Session.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

Affected versions of the Flask package are vulnerable to Information Disclosure due to missing cache-variation headers when the session object is accessed via certain code paths. In Flask’s session handling, accessing flask.session is intended to set a Vary: Cookie response header, but session key-only access patterns (such as using the Python in operator to test for a key without reading or mutating session values) can bypass the logic that adds the header.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Affected versions of gevent were potentially vulnerable to HTTP request smuggling. The issue existed in the pywsgi Input._send_100_continue handling.

Affected versions of gevent are potentially vulnerable to a Race Condition leading to Unauthorized Access — CWE-362.

An issue in Gevent before version 23.9.0 allows a remote attacker to escalate privileges via a crafted script to the WSGIServer component.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in the safe_join function. In Werkzeug versions before 3.1.4, safe_join permits path segments such as “CON” or “AUX” to pass validation, allowing send_from_directory to construct a path that resolves to a Windows device file, which opens successfully but then blocks indefinitely when read.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of the Werkzeug package are vulnerable to Denial of Service (DoS) due to improper handling of Windows special device names in path joining logic. The safe_join() function fails to reject device-name path segments when they are preceded by other segments (for example, example/NUL), and send_from_directory() relies on safe_join() when resolving user-supplied file paths.

Affected versions of the Werkzeug package are vulnerable to Improper Handling of Windows Device Names due to incomplete validation of Windows reserved device names in user-controlled path segments. In werkzeug.security.safe_join, path components ending with special device names (for example CON or AUX) are not reliably rejected when they include compound extensions (for example CON.txt.html) or trailing spaces, and werkzeug.utils.send_from_directory relies on safe_join when resolving a requested filename.

Affected versions of the gunicorn package are vulnerable to HTTP Request/Response Smuggling due to improper validation of the Transfer-Encoding header that enables a TE.CL desynchronisation condition. Gunicorn’s HTTP parser does not consistently enforce RFC semantics when parsing conflicting or unsupported request framing—such as messages containing both Transfer-Encoding and Content-Length—so the backend may fall back to Content-Length while an intermediary treats the message as chunked.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.