GitHub advisories — pip

20 items in the render window (newest 100 + all corrections; not this source's all-time total) · status healthy · last successful fetch 2026-11-21 19:16 UTC
Feed for this sourceRSS
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-21 19:19 UTC

AIT-GUI route chain allows unauthenticated Python execution

Do:Upgrade ait-gui to version 2.5.1 or later.
WHAT CHANGEDGitHub published GHSA-jrjx-9jjw-747m on 20 November 2026, rating CVE-2026-77309 high severity. The advisory states that a route chain in ait-gui allows unauthenticated Python execution in the server process. Version 2.5.1 patches the vulnerability.
WHY IT MATTERSAn unauthenticated attacker can execute arbitrary Python code in the server process, achieving full remote code execution without credentials. Any exposed ait-gui instance running an unpatched version is compromised on contact.
WHAT TO DOUpgrade ait-gui to version 2.5.1 or later. If you cannot upgrade immediately, take the service offline or restrict network access to authenticated internal users only until patched.
Model-reported confidence 95% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-jrjx-9jjw-747m
+summary: AIT-GUI route chain allows unauthenticated Python execution in server process
+severity: high
+published_at: 2026-11-20T23:36:41Z
+cve_id: CVE-2026-77309
+vulnerabilities[].package.name: ait-gui
+vulnerabilities[].first_patched_version: 2.5.1
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningUnauthenticated RCE in a Python package that runs as a server. Platform leads patch this week.
signalsunauthenticated remote code execution in server process · high severity rating · Python package in dependency tree · new advisory with fixed version available · affects a web service (route chain, server process)
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-20 07:21 UTC

Mistune 3.3.3 and earlier: quadratic-time DoS in inline image parsing

Do:Check whether mistune is in your dependency tree and whether you are running 3.3.3 or earlier.
WHAT CHANGEDGitHub published GHSA-ww7q-mxrj-5ffg on 19 November 2026, rating a denial-of-service vulnerability in mistune (CVE-2026-97699) as high severity. The flaw is in inline image parsing (precedence_scan) and bypasses not-is-image guards. First patched version: 3.3.4.
WHY IT MATTERSMistune is a markdown parser in Python dependency trees. An attacker can trigger quadratic-time processing, causing the parser to hang or exhaust resources on crafted input.
WHAT TO DOCheck whether mistune is in your dependency tree and whether you are running 3.3.3 or earlier. If so, update to 3.3.4.
Model-reported confidence 95% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-ww7q-mxrj-5ffg
+summary: Mistune: Quadratic-time DoS in inline image parsing (precedence_scan) bypasses not-is-image guards
+severity: high
+published_at: 2026-11-19T22:44:00Z
+cve_id: CVE-2026-97699
+vulnerabilities[].package.name: mistune
+vulnerabilities[].first_patched_version: 3.3.4
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningMistune is a widely-used Python markdown parser. DoS in parsing logic affects services processing untrusted markdown.
signalsnew advisory published · high severity rating · known CVE assigned · mistune is a markdown parser in Python dependency tree · first patched version specified (3.3.4) · DoS vulnerability affecting parsing logic
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-20 07:21 UTC

MCP Python SDK could send OAuth credentials to attacker-chosen server

Do:Check whether you depend on mcp (Model Context Protocol Python SDK) at any version below 1.30.0 or 2.x below 2.2.0.
WHAT CHANGEDGitHub published GHSA-qx49-fqc8-xw99 (CVE-2026-115563) on 19 November 2026, rating it high severity. The Model Context Protocol Python SDK mcp package allows an MCP server to choose the OAuth authorization server, enabling credential interception. Patched in mcp 1.30.0 and 2.2.0.
WHY IT MATTERSAn application using the vulnerable SDK to authenticate with an MCP server can be tricked into sending OAuth credentials to an attacker-controlled authorization endpoint. The attacker gains the credentials and can impersonate the application.
WHAT TO DOCheck whether you depend on mcp (Model Context Protocol Python SDK) at any version below 1.30.0 or 2.x below 2.2.0. If so, update to 1.30.0, 2.2.0 or later in this cycle.
Model-reported confidence 95% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-qx49-fqc8-xw99
+summary: Model Context Protocol Python SDK: OAuth client could send credentials to an authorization server chosen by the MCP server
+severity: high
+published_at: 2026-11-19T22:45:40Z
+cve_id: CVE-2026-115563
+vulnerabilities[].package.name: mcp, mcp
+vulnerabilities[].first_patched_version: 2.2.0, 1.30.0
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningOAuth credential leakage in widely-used AI SDK; patched versions available; security and platform leads run MCP clients.
signalsAI provider SDK affected (Model Context Protocol Python SDK) · OAuth credential handling vulnerability · High severity rating · New advisory with CVE and patched versions identified · Affects mcp package in dependency tree
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-20 07:21 UTC

Mistune 3.3.3 and earlier vulnerable to quadratic-time DoS

Do:Upgrade mistune to 3.3.4.
WHAT CHANGEDGitHub published GHSA-phmp-hrw4-r36r on 19 November 2026, rating CVE-2026-97698 high severity. Mistune versions before 3.3.4 are vulnerable to quadratic-time denial of service through unbounded parenthesis scanning when parsing empty-label links. Version 3.3.4 patches the vulnerability.
WHY IT MATTERSAn attacker can cause CPU exhaustion in any service that parses untrusted Markdown through mistune, including user-generated content, documentation sites, and API endpoints that accept Markdown input. The quadratic complexity means processing time grows exponentially with input size.
WHAT TO DOUpgrade mistune to 3.3.4. Check your dependency tree with `pip list | grep mistune` or `poetry show mistune` to confirm the installed version, as mistune is a transitive dependency in many Python Markdown toolchains.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-phmp-hrw4-r36r
+summary: Mistune: Quadratic-time DoS in parse_link_href via unbounded parenthesis scan on empty-label links
+severity: high
+published_at: 2026-11-19T22:44:42Z
+cve_id: CVE-2026-97698
+vulnerabilities[].package.name: mistune
+vulnerabilities[].first_patched_version: 3.3.4
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.78, needed 0.5
triage reasoningMistune is a widely-used Python markdown parser. DoS vulnerability with patch available warrants evaluation.
signalsnew advisory published · high severity rating · known CVE identifier assigned · patched version available (3.3.4) · mistune is a markdown parser in Python dependency trees
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-20 07:21 UTC

NLTK 3.10.3 and earlier allows local code execution via prover9/mace4 config

Do:Upgrade nltk to 3.10.4 or later.
WHAT CHANGEDGitHub published GHSA-cc5r-64rf-75hg on 19 November 2026, rating CVE-2026-129234 high severity. NLTK versions before 3.10.4 use an untrusted search path in prover9/mace4 configuration, allowing local code execution through a relative binary_location value. Version 3.10.4 patches the vulnerability.
WHY IT MATTERSAn attacker with local access can execute arbitrary code by placing a malicious binary in the search path before NLTK resolves the prover9 or mace4 binary location. This affects any deployment where NLTK's theorem proving components run with untrusted local users or in shared environments.
WHAT TO DOUpgrade nltk to 3.10.4 or later. If you cannot upgrade immediately, verify whether your code calls NLTK's prover9 or mace4 interfaces and restrict local access to systems running those components.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-cc5r-64rf-75hg
+summary: NLTK: Untrusted search path in prover9/mace4 config allows local code execution via relative binary_location
+severity: high
+published_at: 2026-11-19T23:46:54Z
+cve_id: CVE-2026-129234
+vulnerabilities[].package.name: nltk
+vulnerabilities[].first_patched_version: 3.10.4
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningPython package with local code execution via untrusted search path. Affects dependency trees.
signalsPython package (nltk) in typical dependency trees · Local code execution vulnerability · First patched version specified (3.10.4) · High severity rating · Known CVE assigned
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-20 07:20 UTC

django-docutils before 0.30.0 allows file disclosure, SSRF, and XSS

Do:Upgrade django-docutils to 0.30.0 or later.
WHAT CHANGEDGitHub published GHSA-86mf-q4h5-69gp (CVE-2026-59988) on 20 November 2026, rating django-docutils high severity. Versions before 0.30.0 use unsafe Docutils defaults that enable local file disclosure, server-side request forgery, and stored cross-site scripting. Version 0.30.0 patches the vulnerability.
WHY IT MATTERSApplications using django-docutils to render user-supplied reStructuredText can leak local files to attackers, make requests to internal services, or store malicious scripts that execute in other users' browsers. All three attack vectors are exploitable through the same unsafe configuration.
WHAT TO DOUpgrade django-docutils to 0.30.0 or later. Check whether your application accepts reStructuredText input from users and passes it to django-docutils rendering functions.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-86mf-q4h5-69gp
+summary: django-docutils unsafe Docutils defaults allow local file disclosure, SSRF, and stored XSS
+severity: high
+published_at: 2026-11-20T00:08:12Z
+cve_id: CVE-2026-59988
+vulnerabilities[].package.name: django-docutils
+vulnerabilities[].first_patched_version: 0.30.0
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningDjango-related package with high-severity vulnerability affecting web frameworks commonly in production dependency trees.
signalsnew advisory published · high severity rating · Django package (web framework in dependency tree) · multiple attack vectors: local file disclosure, SSRF, stored XSS · patched version available (0.30.0)
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITYAPI GitHub advisories — pip · 2026-11-20 07:20 UTC

flyto-core 2.26.8 patches SSRF in http.paginate Link header handling

Do:Upgrade flyto-core to 2.26.8 or later.
WHAT CHANGEDCVE-2026-77315 published 20 November 2026. flyto-core's http.paginate follows server-supplied next-page URLs from Link headers or cursors without revalidating the destination, enabling SSRF. First patched version is 2.26.8.
WHY IT MATTERSAn attacker controlling a paginated API response can redirect http.paginate to internal endpoints, cloud metadata services, or other restricted targets your application can reach but should not query on external instruction.
WHAT TO DOUpgrade flyto-core to 2.26.8 or later. If you use http.paginate with untrusted APIs, verify you are running a vulnerable version before 2.26.8 and prioritize the update.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-2mr3-rxrq-238c
+summary: Flyto2 Core: http.paginate follows server-supplied next-page URLs (Link header / cursor) without SSRF revalidation
+severity: high
+published_at: 2026-11-20T00:10:14Z
+cve_id: CVE-2026-77315
+vulnerabilities[].package.name: flyto-core
+vulnerabilities[].first_patched_version: 2.26.8
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.78, needed 0.5
triage reasoningSSRF vulnerability in HTTP pagination affecting a pip package; platform leads patch dependency vulnerabilities.
signalsknown exploitation not stated, but SSRF in pagination is a real attack vector · high severity rating · affects a package in dependency trees (flyto-core) · first patched version exists (2.26.8) · new advisory with CVE assignment
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITYAPI GitHub advisories — pip · 2026-11-20 07:20 UTC

Jupyter MCP Server unauthenticated routes allow backend hijack in standalone HTTP mode

Do:Update jupyter-mcp-server to version 1.0.3.
WHAT CHANGEDGitHub published GHSA-24pw-rf5p-mgp7 (CVE-2026-77318) on 20 November 2026, rating it high severity. Jupyter MCP Server before version 1.0.3 exposes unauthenticated management routes that can hijack the Jupyter backend when deployed in standalone HTTP mode.
WHY IT MATTERSAn attacker with network access to the management routes can take control of the Jupyter backend without authentication, compromising notebook execution environments and any data or credentials accessible to the runtime.
WHAT TO DOUpdate jupyter-mcp-server to version 1.0.3. If you run Jupyter MCP Server in standalone HTTP mode, verify whether management routes are exposed to untrusted networks and prioritize the update.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-24pw-rf5p-mgp7
+summary: Jupyter MCP Server: Unauthenticated management routes can hijack the Jupyter backend in standalone HTTP mode
+severity: high
+published_at: 2026-11-20T00:12:14Z
+cve_id: CVE-2026-77318
+vulnerabilities[].package.name: jupyter-mcp-server
+vulnerabilities[].first_patched_version: 1.0.3
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningUnauthenticated hijack of Jupyter backend affects developer infrastructure and CI/CD pipelines that run notebooks.
signalsnew advisory published · high severity rating · unauthenticated management routes · backend hijack capability · affects Jupyter ecosystem (developer tool / notebook runtime) · patched version available (1.0.3) · standalone HTTP mode deployment vector
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-17 07:57 UTC

mail-parser 4.6.0 and earlier vulnerable to ReDoS in header parsing

Do:Check whether you run mail-parser 4.6.0 or earlier.
WHAT CHANGEDGitHub published GHSA-r7w7-vpq8-c5vv on 16 November 2026, rating mail-parser vulnerable to polynomial regular expression denial of service (ReDoS) during email header parsing. CVE-2026-129295 assigned. First patched version is 4.6.1.
WHY IT MATTERSAn attacker can cause CPU exhaustion by sending crafted email headers, blocking the parser and any service that depends on it. Services that parse untrusted email are exposed.
WHAT TO DOCheck whether you run mail-parser 4.6.0 or earlier. If so, update to 4.6.1.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-r7w7-vpq8-c5vv
+summary: mail-parser is vulnerable to polynomial ReDoS in email header parsing
+severity: high
+published_at: 2026-11-16T22:11:08Z
+cve_id: CVE-2026-129295
+vulnerabilities[].package.name: mail-parser
+vulnerabilities[].first_patched_version: 4.6.1
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.78, needed 0.5
triage reasoningEmail parsing is a common dependency in web services and backend systems. High-severity ReDoS in header parsing affects availability.
signalsnew advisory published · high severity rating · known CVE assigned · first patched version identified · mail-parser is a dependency in many services
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-17 07:57 UTC

mail-parser 4.6.2 patches sender IP spoofing in get_server_ipaddress()

Do:Check whether mail-parser is in your dependency tree and upgrade to 4.6.2 or later if you are running an affected version.
WHAT CHANGEDGitHub published GHSA-72px-j8gh-p79x on 16 November 2026, a high-severity advisory for CVE-2026-129297 in mail-parser. The vulnerability allows sender IP address spoofing in the get_server_ipaddress() function. Version 4.6.2 patches the issue.
WHY IT MATTERSApplications using mail-parser to extract or validate sender IP addresses can be deceived about message origin, bypassing IP-based access controls, allowlists, or audit trails that depend on accurate sender identification.
WHAT TO DOCheck whether mail-parser is in your dependency tree and upgrade to 4.6.2 or later if you are running an affected version.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-72px-j8gh-p79x
+summary: mail-parser: Sender IP address spoofing in get_server_ipaddress()
+severity: high
+published_at: 2026-11-16T22:11:54Z
+cve_id: CVE-2026-129297
+vulnerabilities[].package.name: mail-parser
+vulnerabilities[].first_patched_version: 4.6.2
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.72, needed 0.5
triage reasoningEmail parsing library with IP spoofing vulnerability affecting authentication and logging. Likely in service dependency trees.
signalsnew advisory published · high severity rating · sender IP spoofing vulnerability · patched version available (4.6.2) · mail-parser is email processing library in dependency trees
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-16 22:03 UTC

aiohttp 3.14.4 patches DoS via URL-encoded form fields

Do:Upgrade aiohttp to 3.14.4 or later.
WHAT CHANGEDGitHub published GHSA-xgw8-8ccx-rmqh on 16 November 2026, rating CVE-2026-128126 high severity. The vulnerability affects aiohttp versions before 3.14.4: too many URL-encoded form fields block the event loop, causing denial of service. Version 3.14.4 patches the issue.
WHY IT MATTERSAn attacker can block the event loop by sending requests with excessive URL-encoded form fields, making the service unresponsive. Any Python service using aiohttp to handle untrusted HTTP requests is exposed.
WHAT TO DOUpgrade aiohttp to 3.14.4 or later. Check requirements files and deployed containers for aiohttp < 3.14.4.
Model-reported confidence 95% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-xgw8-8ccx-rmqh
+summary: AIOHTTP: Too many URL-encoded form fields block the event loop
+severity: high
+published_at: 2026-11-16T14:54:10Z
+cve_id: CVE-2026-128126
+vulnerabilities[].package.name: aiohttp
+vulnerabilities[].first_patched_version: 3.14.4
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningaiohttp is a standard Python async HTTP library; event loop DoS affects service availability and warrants patch evaluation.
signalsnew advisory published · aiohttp affected — Python async web framework in typical service dependency trees · high severity rating · patched version available (3.14.4) · DoS vector (event loop blocking) — operational impact
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-16 22:03 UTC

aiohttp unbounded buffering in upgrade requests

Do:Check your dependency tree for aiohttp versions below 3.14.4.
WHAT CHANGEDGitHub published GHSA-pw86-h8fm-j323 (CVE-2026-128129) on 16 November 2026, rating a high-severity unbounded buffering vulnerability in aiohttp during upgrade request handling. First patched version is 3.14.4.
WHY IT MATTERSUnbounded buffering during request handling allows an attacker to exhaust memory by sending crafted upgrade requests, causing service degradation or denial of service in applications using affected aiohttp versions.
WHAT TO DOCheck your dependency tree for aiohttp versions below 3.14.4. If present, update to 3.14.4 or later in this cycle.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-pw86-h8fm-j323
+summary: AIOHTTP: Unbounded buffering of data received during an upgrade request
+severity: high
+published_at: 2026-11-16T14:52:05Z
+cve_id: CVE-2026-128129
+vulnerabilities[].package.name: aiohttp
+vulnerabilities[].first_patched_version: 3.14.4
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningaiohttp is a widely-used Python async HTTP library in production services. High-severity vulnerability with patch available.
signalsPython package (aiohttp) in typical service dependency tree · High severity rating · Known CVE assigned · First patched version specified (3.14.4) · Unbounded buffering vulnerability affects request handling
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-16 22:03 UTC

aiohttp 3.14.4 patches event loop stall in wildcard matching

Do:Upgrade aiohttp to 3.14.4 or later.
WHAT CHANGEDGitHub published GHSA-jm3w-gw2p-xjfj on 16 November 2026, a high-severity advisory for CVE-2026-128124 in aiohttp. The vulnerability causes event loop stalls in add_domain() wildcard matching. First patched version is 3.14.4.
WHY IT MATTERSAn attacker can stall the async event loop, blocking all concurrent requests and responses in the application. This is a denial of service vector in the core async runtime that affects any service using aiohttp's domain matching.
WHAT TO DOUpgrade aiohttp to 3.14.4 or later. Check requirements files and lock files for aiohttp dependencies below 3.14.4.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-jm3w-gw2p-xjfj
+summary: AIOHTTP: Event loop stall in add_domain() wildcard matching
+severity: high
+published_at: 2026-11-16T14:56:35Z
+cve_id: CVE-2026-128124
+vulnerabilities[].package.name: aiohttp
+vulnerabilities[].first_patched_version: 3.14.4
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningaiohttp is a standard Python dependency for async services. Event loop stall is actionable DoS risk.
signalsnew advisory published · aiohttp affected — widely-used Python async HTTP library in dependency trees · high severity rating · patched version available (3.14.4) · event loop stall — denial of service vector in core async runtime
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-16 22:03 UTC

aiohttp 3.14.4 patches DoS via concatenated compressed streams

Do:Upgrade aiohttp to 3.14.4 or later.
WHAT CHANGEDGitHub published GHSA-h4cc-hm5h-fmm9 (CVE-2026-128131) on 16 November 2026, rating it high severity. aiohttp versions before 3.14.4 consume excessive CPU when processing concatenated compressed streams, creating a denial-of-service vector.
WHY IT MATTERSServices using aiohttp to handle HTTP requests can be forced into CPU exhaustion by a crafted payload, degrading or stopping availability. This affects any deployment where aiohttp processes untrusted input.
WHAT TO DOUpgrade aiohttp to 3.14.4 or later. Grep dependency manifests for aiohttp and check running versions against the patched release.
Model-reported confidence 95% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-h4cc-hm5h-fmm9
+summary: AIOHTTP: Concatenated compressed streams cause excessive CPU use
+severity: high
+published_at: 2026-11-16T14:51:27Z
+cve_id: CVE-2026-128131
+vulnerabilities[].package.name: aiohttp
+vulnerabilities[].first_patched_version: 3.14.4
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningaiohttp is a standard Python HTTP library; high-severity DoS in a runtime dependency warrants patching review.
signalsknown advisory with CVE identifier · high severity rating · aiohttp is a widely-used Python async HTTP library in dependency trees · first patched version specified (3.14.4) · DoS vector (excessive CPU use) affects availability of services running aiohttp
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-16 22:02 UTC

NLTK dot2img enables local code execution via PATH manipulation

Do:Upgrade nltk to 3.10.4 or later.
WHAT CHANGEDCVE-2026-128134 published 16 November 2026. NLTK's dot2img function runs Graphviz 'dot' without the trusted-exec chokepoint, allowing local code execution when an attacker controls a writable PATH element. First patched version: 3.10.4.
WHY IT MATTERSAn attacker with write access to any directory in PATH can place a malicious 'dot' executable that NLTK will run with the application's privileges when dot2img is called. This bypasses the intended security boundary for external tool invocation.
WHAT TO DOUpgrade nltk to 3.10.4 or later. If you cannot upgrade immediately, audit PATH for writable directories and remove write permissions for untrusted users.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-7mxv-7h3q-9324
+summary: NLTK dot2img runs Graphviz 'dot' without the trusted-exec chokepoint, enabling local code execution via a writable PATH element
+severity: high
+published_at: 2026-11-16T18:45:51Z
+cve_id: CVE-2026-128134
+vulnerabilities[].package.name: nltk
+vulnerabilities[].first_patched_version: 3.10.4
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningNLTK is a widely-used Python NLP library in data science and ML pipelines. Local code execution via PATH is actionable.
signalsknown exploitation vector: local code execution via PATH manipulation · high severity rating · affects Python package in typical dependency trees · first patched version specified (3.10.4) · new advisory with published date
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-16 22:02 UTC

aiohttp WebSocket receive queue limits bypassed by small messages

Do:Upgrade aiohttp to 3.14.4 or later.
WHAT CHANGEDCVE-2026-128132 published 16 November 2026 as GHSA-5rmc-48h2-hvcm, rated high severity. Small or compressed WebSocket messages bypass receive queue limits in aiohttp. First patched version is 3.14.4.
WHY IT MATTERSAn attacker can exhaust server resources by sending WebSocket messages that evade queue limits, causing denial of service. Applications using aiohttp WebSocket handling are exposed to resource exhaustion attacks.
WHAT TO DOUpgrade aiohttp to 3.14.4 or later. Grep for 'aiohttp' in requirements files and check deployed versions against the CVE scope.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-5rmc-48h2-hvcm
+summary: AIOHTTP: WebSocket receive queue limits bypassed by small or compressed messages
+severity: high
+published_at: 2026-11-16T14:51:01Z
+cve_id: CVE-2026-128132
+vulnerabilities[].package.name: aiohttp
+vulnerabilities[].first_patched_version: 3.14.4
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningHigh-severity DoS in aiohttp WebSocket handling. Python runtime library in common service dependency trees.
signalsknown CVE with assigned ID · high severity rating · aiohttp is a widely-used Python async HTTP library in dependency trees · WebSocket handling is core infrastructure code · receive queue limits bypass is a DoS/resource exhaustion vector · patched version available (3.14.4) · new advisory entry (not a restatement)
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-16 22:02 UTC

aiohttp WebSocket integer overflow bypasses max_msg_size

Do:Update aiohttp to 3.14.4 or later.
WHAT CHANGEDCVE-2026-128128 published 16 November 2026 for aiohttp, severity high. An integer overflow in WebSocket size validation allows messages to bypass max_msg_size limits. Patched in version 3.14.4.
WHY IT MATTERSServices using aiohttp WebSockets can receive messages larger than configured limits, potentially causing memory exhaustion or denial of service. The bypass is in protocol-level validation, so application-layer size checks do not prevent it.
WHAT TO DOUpdate aiohttp to 3.14.4 or later. If you run WebSocket endpoints with aiohttp, prioritize this in the current patch cycle.
Model-reported confidence 95% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-3rp8-rxj8-mvjw
+summary: AIOHTTP: WebSocket max_msg_size bypass via integer overflow
+severity: high
+published_at: 2026-11-16T14:52:40Z
+cve_id: CVE-2026-128128
+vulnerabilities[].package.name: aiohttp
+vulnerabilities[].first_patched_version: 3.14.4
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningaiohttp is a standard Python async HTTP library; WebSocket max_msg_size bypass affects services running it.
signalsknown CVE with assigned ID · high severity rating · affects aiohttp — Python async HTTP library in typical service dependency trees · WebSocket protocol handling vulnerability (integer overflow in size validation) · patched version available (3.14.4) · new advisory entry — not a restatement
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
ACTION NEEDED SECURITY GitHub advisories — pip · 2026-11-16 22:02 UTC

aiohttp 3.14.4 patches request replay and buffer growth vulnerability

Do:Check your dependency tree for aiohttp versions below 3.14.4.
WHAT CHANGEDGitHub published CVE-2026-128130 on 16 November 2026, a high-severity vulnerability in aiohttp's pure-Python parser that allows request replay and buffer growth behind a declined upgrade. First patched version is 3.14.4.
WHY IT MATTERSRequest replay can bypass authentication or rate limiting by re-executing requests the server already declined. Unbounded buffer growth consumes memory until the process fails or the host runs out of resources.
WHAT TO DOCheck your dependency tree for aiohttp versions below 3.14.4. If present, update to 3.14.4 or later in this cycle.
Model-reported confidence 85% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-28qf-cpwq-vmc2
+summary: AIOHTTP: Pure-Python parser request replay and buffer growth behind a declined upgrade
+severity: high
+published_at: 2026-11-16T14:51:53Z
+cve_id: CVE-2026-128130
+vulnerabilities[].package.name: aiohttp
+vulnerabilities[].first_patched_version: 3.14.4
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningaiohttp is a standard Python runtime dependency; high-severity advisory with patch available warrants platform lead action.
signalsaiohttp is a widely-used Python async HTTP client/server library in dependency trees · high severity rating · known CVE assigned · first patched version specified (3.14.4) · request replay and buffer growth are memory safety and request handling concerns
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)
Corrections and arithmetic notes
CORRECTED dnspython CVE-2026-128089 causes excessive memory and CPU use
CORRECTED 2026-11-21 15:59 UTC — Correction, 21 November 2026: This item said services using dnspython for DNS resolution can be denied service, and told readers to upgrade wherever the package is a runtime dependency. The advisory is narrower: the from_text() parser can consume excessive memory and CPU when it reads a WKS record in attacker-controlled zone-file text, because it does not limit the service number to 0–65535. Applications are affected only when they parse zone-file text whose record type and rdata are attacker-controlled. The high severity rating and the patched version, 2.9.2, were stated correctly. GitHub's Advisory Database publication date of 13 November 2026 was also correct; the project advisory itself was published on 4 November 2026.
GitHub advisories — pip · published 2026-11-15 06:36 UTC · corrected 2026-11-21 15:59 UTC · stood 153h 23m · cause writer overreach
WHAT CHANGED (CORRECTED)GitHub published GHSA-vvf4-9x63-8xvv on 13 November 2026, rating CVE-2026-128089 in dnspython as high severity. The vulnerability is in zone-file parsing: the `from_text()` parser does not limit a WKS record service number to 0–65535, so a maliciously crafted zone file can cause excessive memory and CPU consumption. Versions earlier than 2.9.2 are affected. First patched version is 2.9.2.
Original text (as first published, superseded)
WHAT CHANGEDGitHub published GHSA-vvf4-9x63-8xvv on 13 November 2026, rating CVE-2026-128089 in dnspython as high severity. The vulnerability causes excessive memory and CPU consumption. First patched version is 2.9.2.
WHY IT MATTERSServices using dnspython for DNS resolution can experience availability degradation or denial of service through resource exhaustion. The high severity rating indicates exploitability affects production workloads.
WHAT TO DOUpgrade dnspython to version 2.9.2 or later in services where it is a runtime dependency.
TRIAGE SIGNALDoS vector (excessive memory and CPU use) affects availability
TRIAGE REASONINGdnspython is a common DNS library in Python services. High-severity DoS advisory with patch available warrants review.
Full item and evidence ↗
CORRECTED PyJWT 2.13.0 accepts JWK public keys as HMAC secrets
CORRECTED 2026-10-18 15:00 UTC — Correction: The affected range was overstated. GHSA-w2cx-738m-mc7w lists only PyJWT = 2.13.0 as vulnerable, not every version before 2.14.0. Patched in 2.14.0 is correct.
GitHub advisories — pip · published 2026-09-30 21:59 UTC · corrected 2026-10-18 15:00 UTC · stood 425h 0m · cause writer overreach
WHAT CHANGED (CORRECTED)CVE-2026-102273 (GHSA-w2cx-738m-mc7w) published 29 September 2026, rated high severity. PyJWT 2.13.0 accepts public JWK containers as HMAC secrets. Only version 2.13.0 is listed as vulnerable. Patched in 2.14.0.
Original text (as first published, superseded)
HEADLINEPyJWT accepts JWK public keys as HMAC secrets
WHAT CHANGEDCVE-2026-102273 published 29 September 2026. PyJWT before 2.14.0 accepts public JWK containers as HMAC secrets, rated high severity. Patched in 2.14.0.
WHY IT MATTERSAn attacker who controls the JWK input can forge tokens by supplying a public key where the application expects an HMAC secret, bypassing signature verification entirely. Any JWT validation using HMAC with externally-supplied keys is compromised.
WHAT TO DOGrep your dependency trees for PyJWT <2.14.0. If found, update to 2.14.0 this cycle. If you pass JWK data into JWT verification, audit those call sites for untrusted input before deploying the patch.
Full item and evidence ↗