<?xml version='1.0' encoding='UTF-8'?>
<?xml-stylesheet href="/static/style.xsl" type="text/xsl"?>
<rss xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/" version="2.0">
  <channel>
    <title>Most recent entries from pysec</title>
    <link>https://vulnerability.circl.lu</link>
    <description>Contains only the most 10 recent entries.</description>
    <docs>http://www.rssboard.org/rss-specification</docs>
    <generator>python-feedgen</generator>
    <language>en</language>
    <lastBuildDate>Sun, 22 Nov 2026 14:17:06 +0000</lastBuildDate>
    <item>
      <title>PYSEC-2025-19</title>
      <link>https://vulnerability.circl.lu/vuln/pysec-2025-19</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: picklescan&lt;/p&gt;
&lt;p&gt;picklescan before 0.0.22 only considers standard pickle file extensions in the scope for its vulnerability scan. An attacker could craft a malicious model that uses Pickle and include a malicious pickle file with a non-standard file extension. Because the malicious pickle file inclusion is not considered as part of the scope of picklescan, the file would pass security checks and appear to be safe, when it could instead prove to be problematic.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: picklescan&lt;/p&gt;
&lt;p&gt;picklescan before 0.0.22 only considers standard pickle file extensions in the scope for its vulnerability scan. An attacker could craft a malicious model that uses Pickle and include a malicious pickle file with a non-standard file extension. Because the malicious pickle file inclusion is not considered as part of the scope of picklescan, the file would pass security checks and appear to be safe, when it could instead prove to be problematic.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://vulnerability.circl.lu/vuln/pysec-2025-19</guid>
      <pubDate>Mon, 03 Mar 2025 19:15:34 +0000</pubDate>
    </item>
    <item>
      <title>PYSEC-2024-115</title>
      <link>https://vulnerability.circl.lu/vuln/pysec-2024-115</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: langchain-community&lt;/p&gt;
&lt;p&gt;A vulnerability in the GraphCypherQAChain class of langchain-ai/langchain-community version 0.2.5 allows for SQL injection through prompt injection. This vulnerability can lead to unauthorized data manipulation, data exfiltration, denial of service (DoS) by deleting all data, breaches in multi-tenant security environments, and data integrity issues. Attackers can create, update, or delete nodes and relationships without proper authorization, extract sensitive data, disrupt services, access data across different tenants, and compromise the integrity of the database.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: langchain-community&lt;/p&gt;
&lt;p&gt;A vulnerability in the GraphCypherQAChain class of langchain-ai/langchain-community version 0.2.5 allows for SQL injection through prompt injection. This vulnerability can lead to unauthorized data manipulation, data exfiltration, denial of service (DoS) by deleting all data, breaches in multi-tenant security environments, and data integrity issues. Attackers can create, update, or delete nodes and relationships without proper authorization, extract sensitive data, disrupt services, access data across different tenants, and compromise the integrity of the database.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://vulnerability.circl.lu/vuln/pysec-2024-115</guid>
      <pubDate>Tue, 05 Nov 2024 16:04:14 +0000</pubDate>
    </item>
    <item>
      <title>PYSEC-2026-2946 — PraisonAI: Coarse-Grained Tool Approval Cache Bypasses Per-Invocation Consent for Shell Commands</title>
      <link>https://vulnerability.circl.lu/vuln/pysec-2026-2946</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: praisonaiagents&lt;/p&gt;
&lt;p&gt;## Summary&lt;/p&gt;
&lt;p&gt;The approval system in PraisonAI Agents caches tool approval decisions by tool name only, not by invocation arguments. Once a user approves `execute_command` for any command (e.g., `ls -la`), all subsequent `execute_command` calls in that execution context bypass the approval prompt entirely. Combined with `os.environ.copy()` passing all process environment variables to subprocesses, this allows an LLM agent (potentially via prompt injection) to silently exfiltrate API keys and credentials without further user consent.&lt;/p&gt;
&lt;p&gt;## Details&lt;/p&gt;
&lt;p&gt;The `require_approval` decorator in `src/praisonai-agents/praisonaiagents/approval/__init__.py:176-178` checks approval status by tool name only:&lt;/p&gt;
&lt;p&gt;```python
@wraps(func)
def wrapper(*args, **kwargs):
    if is_already_approved(tool_name):   # line 177 — checks only tool_name
        return func(*args, **kwargs)     # line 178 — bypasses ALL approval
```&lt;/p&gt;
&lt;p&gt;The `mark_approved` function in `registry.py:144-147` stores only the tool name string:&lt;/p&gt;
&lt;p&gt;```python
def mark_approved(self, tool_name: str) -&amp;gt; None:
    approved = self._approved_context.get(set())
    approved.add(tool_name)              # stores &amp;#34;execute_command&amp;#34;, not args
    self._approved_context.set(approved)
```&lt;/p&gt;
&lt;p&gt;The approval context is never cleared during agent execution — `clear_approved()` exists (`registry.py:152`) but is never called in the agent&amp;#39;s tool execution path (`agent/tool_execution.py`).&lt;/p&gt;
&lt;p&gt;Meanwhile, the `ConsoleBackend` UI at `backends.py:95-96` misleads the user:…&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: praisonaiagents&lt;/p&gt;
&lt;p&gt;## Summary&lt;/p&gt;
&lt;p&gt;The approval system in PraisonAI Agents caches tool approval decisions by tool name only, not by invocation arguments. Once a user approves `execute_command` for any command (e.g., `ls -la`), all subsequent `execute_command` calls in that execution context bypass the approval prompt entirely. Combined with `os.environ.copy()` passing all process environment variables to subprocesses, this allows an LLM agent (potentially via prompt injection) to silently exfiltrate API keys and credentials without further user consent.&lt;/p&gt;
&lt;p&gt;## Details&lt;/p&gt;
&lt;p&gt;The `require_approval` decorator in `src/praisonai-agents/praisonaiagents/approval/__init__.py:176-178` checks approval status by tool name only:&lt;/p&gt;
&lt;p&gt;```python
@wraps(func)
def wrapper(*args, **kwargs):
    if is_already_approved(tool_name):   # line 177 — checks only tool_name
        return func(*args, **kwargs)     # line 178 — bypasses ALL approval
```&lt;/p&gt;
&lt;p&gt;The `mark_approved` function in `registry.py:144-147` stores only the tool name string:&lt;/p&gt;
&lt;p&gt;```python
def mark_approved(self, tool_name: str) -&amp;gt; None:
    approved = self._approved_context.get(set())
    approved.add(tool_name)              # stores &amp;#34;execute_command&amp;#34;, not args
    self._approved_context.set(approved)
```&lt;/p&gt;
&lt;p&gt;The approval context is never cleared during agent execution — `clear_approved()` exists (`registry.py:152`) but is never called in the agent&amp;#39;s tool execution path (`agent/tool_execution.py`).&lt;/p&gt;
&lt;p&gt;Meanwhile, the `ConsoleBackend` UI at `backends.py:95-96` misleads the user:…&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://vulnerability.circl.lu/vuln/pysec-2026-2946</guid>
      <pubDate>Mon, 13 Jul 2026 14:36:52 +0000</pubDate>
    </item>
    <item>
      <title>PYSEC-2026-4395</title>
      <link>https://vulnerability.circl.lu/vuln/pysec-2026-4395</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: aiohttp&lt;/p&gt;
&lt;p&gt;AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to 3.14.4, the C WebSocket reader adds a declared frame length to the size of an already buffered partial message using signed 64-bit arithmetic. A length near two to the power of 63 can overflow, allowing the max_msg_size check to pass. A remote peer can send a small non-final fragment followed by an oversized message and cause unbounded memory consumption. The Python WebSocket reader is unaffected. This issue is fixed in version 3.14.4.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: aiohttp&lt;/p&gt;
&lt;p&gt;AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to 3.14.4, the C WebSocket reader adds a declared frame length to the size of an already buffered partial message using signed 64-bit arithmetic. A length near two to the power of 63 can overflow, allowing the max_msg_size check to pass. A remote peer can send a small non-final fragment followed by an oversized message and cause unbounded memory consumption. The Python WebSocket reader is unaffected. This issue is fixed in version 3.14.4.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://vulnerability.circl.lu/vuln/pysec-2026-4395</guid>
      <pubDate>Fri, 13 Nov 2026 21:17:44 +0000</pubDate>
    </item>
    <item>
      <title>PYSEC-2026-4394</title>
      <link>https://vulnerability.circl.lu/vuln/pysec-2026-4394</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: aiohttp&lt;/p&gt;
&lt;p&gt;AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to 3.14.4, BodyPartReader.read_chunk() processing a multipart part with Content-Transfer-Encoding set to base64 keeps reading until the chunk length is a multiple of four without enforcing a byte limit. A short Content-Length or a long run of whitespace can cause one call to read the remaining body into memory before the size check or spin on the event loop. A remote attacker can send such a multipart request to a handler that reads multipart data and bypass client_max_size. This issue is fixed in version 3.14.4.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: aiohttp&lt;/p&gt;
&lt;p&gt;AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to 3.14.4, BodyPartReader.read_chunk() processing a multipart part with Content-Transfer-Encoding set to base64 keeps reading until the chunk length is a multiple of four without enforcing a byte limit. A short Content-Length or a long run of whitespace can cause one call to read the remaining body into memory before the size check or spin on the event loop. A remote attacker can send such a multipart request to a handler that reads multipart data and bypass client_max_size. This issue is fixed in version 3.14.4.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://vulnerability.circl.lu/vuln/pysec-2026-4394</guid>
      <pubDate>Fri, 13 Nov 2026 21:17:44 +0000</pubDate>
    </item>
    <item>
      <title>PYSEC-2026-4393</title>
      <link>https://vulnerability.circl.lu/vuln/pysec-2026-4393</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: aiohttp&lt;/p&gt;
&lt;p&gt;AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to 3.14.4, Request.post() does not limit the number of fields in a URL-encoded form body, and inserting many repeated keys into MultiDict takes quadratic time on the event loop. A remote attacker can send a form POST within client_max_size to a handler that calls Request.post() and block the event loop. This issue is fixed in version 3.14.4.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: aiohttp&lt;/p&gt;
&lt;p&gt;AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to 3.14.4, Request.post() does not limit the number of fields in a URL-encoded form body, and inserting many repeated keys into MultiDict takes quadratic time on the event loop. A remote attacker can send a form POST within client_max_size to a handler that calls Request.post() and block the event loop. This issue is fixed in version 3.14.4.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://vulnerability.circl.lu/vuln/pysec-2026-4393</guid>
      <pubDate>Fri, 13 Nov 2026 21:17:44 +0000</pubDate>
    </item>
    <item>
      <title>PYSEC-2026-4392</title>
      <link>https://vulnerability.circl.lu/vuln/pysec-2026-4392</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: aiohttp&lt;/p&gt;
&lt;p&gt;AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to 3.14.4, Set-Cookie parsing treats unrecognized name=value attributes as additional cookies, while CookieJar does not limit the number of cookies stored or sent. A malicious server can make an aiohttp client accumulate a very large cookie set, consume memory and CPU repeatedly across redirects, and construct an excessively large Cookie header for later requests to the same domain. This issue is fixed in version 3.14.4.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: aiohttp&lt;/p&gt;
&lt;p&gt;AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to 3.14.4, Set-Cookie parsing treats unrecognized name=value attributes as additional cookies, while CookieJar does not limit the number of cookies stored or sent. A malicious server can make an aiohttp client accumulate a very large cookie set, consume memory and CPU repeatedly across redirects, and construct an excessively large Cookie header for later requests to the same domain. This issue is fixed in version 3.14.4.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://vulnerability.circl.lu/vuln/pysec-2026-4392</guid>
      <pubDate>Fri, 13 Nov 2026 21:17:44 +0000</pubDate>
    </item>
    <item>
      <title>PYSEC-2026-4391</title>
      <link>https://vulnerability.circl.lu/vuln/pysec-2026-4391</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: aiohttp&lt;/p&gt;
&lt;p&gt;AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to 3.14.4, Application.add_domain() converts every asterisk wildcard in a domain mask into a greedy regular-expression fragment, so a mask containing two or more wildcards can take polynomial time to reject a long Host header that does not match. A remote attacker can repeatedly send such headers to a server configured with a multi-wildcard domain mask and stall the event loop. This issue is fixed in version 3.14.4.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: aiohttp&lt;/p&gt;
&lt;p&gt;AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to 3.14.4, Application.add_domain() converts every asterisk wildcard in a domain mask into a greedy regular-expression fragment, so a mask containing two or more wildcards can take polynomial time to reject a long Host header that does not match. A remote attacker can repeatedly send such headers to a server configured with a multi-wildcard domain mask and stall the event loop. This issue is fixed in version 3.14.4.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://vulnerability.circl.lu/vuln/pysec-2026-4391</guid>
      <pubDate>Fri, 13 Nov 2026 21:17:44 +0000</pubDate>
    </item>
    <item>
      <title>PYSEC-2026-4390</title>
      <link>https://vulnerability.circl.lu/vuln/pysec-2026-4390</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: tensorrt-llm&lt;/p&gt;
&lt;p&gt;NVIDIA TensorRT-LLM for any platform contains a vulnerability in visual gen server, where an attacker could cause an unsafe deserialization by unauthorized zeroMQ deserialization. A successful exploit of this vulnerability might lead to code execution.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: tensorrt-llm&lt;/p&gt;
&lt;p&gt;NVIDIA TensorRT-LLM for any platform contains a vulnerability in visual gen server, where an attacker could cause an unsafe deserialization by unauthorized zeroMQ deserialization. A successful exploit of this vulnerability might lead to code execution.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://vulnerability.circl.lu/vuln/pysec-2026-4390</guid>
      <pubDate>Tue, 14 Jul 2026 21:16:43 +0000</pubDate>
    </item>
    <item>
      <title>PYSEC-2025-102</title>
      <link>https://vulnerability.circl.lu/vuln/pysec-2025-102</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: dagster-ge&lt;/p&gt;
&lt;p&gt;Local File Inclusion in dagster._grpc.impl.get_notebook_data in Dagster 1.10.14 allows attackers with access to the gRPC server to read arbitrary files by supplying path traversal sequences in the notebook_path field of ExternalNotebookData requests, bypassing the intended extension-based check.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: dagster-ge&lt;/p&gt;
&lt;p&gt;Local File Inclusion in dagster._grpc.impl.get_notebook_data in Dagster 1.10.14 allows attackers with access to the gRPC server to read arbitrary files by supplying path traversal sequences in the notebook_path field of ExternalNotebookData requests, bypassing the intended extension-based check.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://vulnerability.circl.lu/vuln/pysec-2025-102</guid>
      <pubDate>Tue, 22 Jul 2025 17:15:33 +0000</pubDate>
    </item>
  </channel>
</rss>
