The "profiling.sampling" module (Python 3.15+) and "asyncio introspection capabilities" (3.14+, "python -m asyncio ps" and "python -m asyncio pstree") features could be used to read and write addresses in a privileged process if that process connected to a malicious or "infected" Python process via the remote debugging feature. This vulnerability requires persistently and repeatedly connecting to the process to be exploited, even after the connecting process crashes with high likelihood due to ASLR.

PraisonAI is a multi-agent teams system. In versions below 4.5.139 of PraisonAI and 1.5.140 of praisonaiagents, the workflow engine is vulnerable to arbitrary command and code execution through untrusted YAML files. When praisonai workflow run <file.yaml> loads a YAML file with type: job, the JobWorkflowExecutor in job_workflow.py processes steps that support run: (shell commands via subprocess.run()), script: (inline Python via exec()), and python: (arbitrary Python script execution)—all without any validation, sandboxing, or user confirmation. The affected code paths include action_run() in workflow.py and _exec_shell(), _exec_inline_python(), and _exec_python_script() in job_workflow.py. An attacker who can supply or influence a workflow YAML file (particularly in CI pipelines, shared repositories, or multi-tenant deployment environments) can achieve full arbitrary command execution on the host system, compromising the machine and any accessible data or credentials. This issue has been fixed in versions 4.5.139 of PraisonAI and 1.5.140 of praisonaiagents.

praison

PraisonAI is a multi-agent teams system. Versions 4.5.138 and below are vulnerable to arbitrary code execution through automatic, unsanitized import of a tools.py file from the current working directory. Components including call.py (import_tools_from_file()), tool_resolver.py (_load_local_tools()), and CLI tool-loading paths blindly import ./tools.py at startup without any validation, sandboxing, or user confirmation. An attacker who can place a malicious tools.py in the directory where PraisonAI is launched (such as through a shared project, cloned repository, or writable workspace) achieves immediate arbitrary Python code execution in the host environment. This compromises the full PraisonAI process, the host system, and any connected data or credentials. This issue has been fixed in version 4.5.139.

praison

MaxKB is an open-source AI assistant for enterprise. In versions 2.7.1 and below, an authenticated user can bypass sandbox result validation and spoof tool execution results by exploiting Python frame introspection to read the wrapper's UUID from its bytecode constants, then writing a forged result directly to file descriptor 1 (bypassing stdout redirection). By calling sys.exit(0), the attacker terminates the wrapper before it prints the legitimate output, causing the MaxKB service to parse and trust the spoofed response as the genuine tool result. This issue has been fixed in version 2.8.0.

maxkb

MaxKB is an open-source AI assistant for enterprise. Versions 2.7.1 and below contain a sandbox escape vulnerability in the ToolExecutor component. By leveraging Python's ctypes library to execute raw system calls, an authenticated attacker with workspace privileges can bypass the LD_PRELOAD-based sandbox.so module to achieve arbitrary code execution via direct kernel system calls, enabling full network exfiltration and container compromise. The library intercepts critical standard system functions such as execve, system, connect, and open. It also intercepts mprotect to prevent PROT_EXEC (executable memory) allocations within the sandboxed Python processes, but pkey_mprotect is not blocked. This issue has been fixed in version 2.8.0.

maxkb

MaxKB is an open-source AI assistant for enterprise. In versions 2.7.1 and below, an incomplete sandbox protection mechanism allows an authenticated user with tool execution privileges to escape the LD_PRELOAD-based sandbox. By env command the attacker can clear the environment variables and drop the sandbox.so hook, leading to unrestricted Remote Code Execution (RCE) and network access. MaxKB restricts untrusted Python code execution via the Tool Debug API by injecting sandbox.so through the LD_PRELOAD environment variable. This intercepts sensitive C library functions (like execve, socket, open) to restrict network and file access. However, a patch allowed the /usr/bin/env utility to be executed by the sandboxed user. When an attacker is permitted to create subprocesses, they can execute the env -i python command. The -i flag instructs env to completely clear all environment variables before running the target program. This effectively drops the LD_PRELOAD environment variable. The newly spawned Python process will therefore execute natively without any sandbox hooks, bypassing all network and file system restrictions. This issue has been fixed in version 2.8.0.

maxkb

A Code Injection and Missing Authentication vulnerability in Google Agent Development Kit (ADK) versions 1.7.0 (and 2.0.0a1) through 1.28.1 (and 2.0.0a2) on Python (OSS), Cloud Run, and GKE allows an unauthenticated remote attacker to execute arbitrary code on the server hosting the ADK instance. This vulnerability was patched in versions 1.28.1 and 2.0.0a2. Customers need to redeploy the upgraded ADK to their production environments. In addition, if they are running ADK Web locally, they also need to upgrade their local instance.

PraisonAI is a multi-agent teams system. Prior to 4.5.128, PraisonAI’s MCP (Model Context Protocol) integration allows spawning background servers via stdio using user-supplied command strings (e.g., MCP("npx -y @smithery/cli ...")). These commands are executed through Python’s subprocess module. By default, the implementation forwards the entire parent process environment to the spawned subprocess. As a result, any MCP command executed in this manner inherits all environment variables from the host process, including sensitive data such as API keys, authentication tokens, and database credentials. This behavior introduces a security risk when untrusted or third-party commands are used. In common scenarios where MCP tools are invoked via package runners such as npx -y, arbitrary code from external or potentially compromised packages may execute with access to these inherited environment variables. This creates a risk of unintended credential exposure and enables potential supply chain attacks through silent exfiltration of secrets. This vulnerability is fixed in 4.5.128.

praison

PraisonAI is a multi-agent teams system. Prior to 4.5.128, PraisonAI's AST-based Python sandbox can be bypassed using type.__getattribute__ trampoline, allowing arbitrary code execution when running untrusted agent code. The _execute_code_direct function in praisonaiagents/tools/python_tools.py uses AST filtering to block dangerous Python attributes like __subclasses__, __globals__, and __bases__. However, the filter only checks ast.Attribute nodes, allowing a bypass. The sandbox relies on AST-based filtering of attribute access but fails to account for dynamic attribute resolution via built-in methods such as type.getattribute, resulting in incomplete enforcement of security restrictions. The string '__subclasses__' is an ast.Constant, not an ast.Attribute, so it is never checked against the blocked list. This vulnerability is fixed in 4.5.128.

praison

PraisonAIAgents is a multi-agent teams system. Prior to 1.5.128, he list_files() tool in FileTools validates the directory parameter against workspace boundaries via _validate_path(), but passes the pattern parameter directly to Path.glob() without any validation. Since Python's Path.glob() supports .. path segments, an attacker can use relative path traversal in the glob pattern to enumerate arbitrary files outside the workspace, obtaining file metadata (existence, name, size, timestamps) for any path on the filesystem. This vulnerability is fixed in 1.5.128.

praison

An Execution with Unnecessary Privileges vulnerability in the User Interface (UI) of Juniper Networks Junos OS and Junos OS Evolved allows a local, low-privileged attacker to gain root privileges, thus compromising the system. When a configuration that allows unsigned Python op scripts is present on the device, a non-root user is able to execute malicious op scripts as a root-equivalent user, leading to privilege escalation.  This issue affects Junos OS:  * All versions before 22.4R3-S7,  * from 23.2 before 23.2R2-S4,  * from 23.4 before 23.4R2-S6, * from 24.2 before 24.2R1-S2, 24.2R2,  * from 24.4 before 24.4R1-S2, 24.4R2;  Junos OS Evolved:  * All versions before 22.4R3-S7-EVO,  * from 23.2 before 23.2R2-S4-EVO,  * from 23.4 before 23.4R2-S6-EVO, * from 24.2 before 24.2R2-EVO,  * from 24.4 before 24.4R1-S1-EVO, 24.4R2-EVO.

juniper

LangChain is a framework for building agents and LLM-powered applications. Prior to 0.3.84 and 1.2.28, LangChain's f-string prompt-template validation was incomplete in two respects. First, some prompt template classes accepted f-string templates and formatted them without enforcing the same attribute-access validation as PromptTemplate. In particular, DictPromptTemplate and ImagePromptTemplate could accept templates containing attribute access or indexing expressions and subsequently evaluate those expressions during formatting. Second, f-string validation based on parsed top-level field names did not reject nested replacement fields inside format specifiers. In this pattern, the nested replacement field appears in the format specifier rather than in the top-level field name. As a result, earlier validation based on parsed field names did not reject the template even though Python formatting would still attempt to resolve the nested expression at runtime. This vulnerability is fixed in 0.3.84 and 1.2.28.

langchain

web3.py allows you to interact with the Ethereum blockchain using Python. From 6.0.0b3 to before 7.15.0 and 8.0.0b2, web3.py implements CCIP Read / OffchainLookup (EIP-3668) by performing HTTP requests to URLs supplied by smart contracts in offchain_lookup_payload["urls"]. The implementation uses these contract-supplied URLs directly (after {sender} / {data} template substitution) without any destination validation. CCIP Read is enabled by default (global_ccip_read_enabled = True on all providers), meaning any application using web3.py's .call() method is exposed without explicit opt-in. This results in Server-Side Request Forgery (SSRF) when web3.py is used in backend services, indexers, APIs, or any environment that performs eth_call / .call() against untrusted or user-supplied contract addresses. A malicious contract can force the web3.py process to issue HTTP requests to arbitrary destinations, including internal network services and cloud metadata endpoints. This vulnerability is fixed in 7.15.0 and 8.0.0b2.

pyLoad is a free and open-source download manager written in Python. Prior to 0.5.0b3.dev97, the /json/package_order, /json/link_order, and /json/abort_link WebUI JSON endpoints enforce weaker permissions than the core API methods they invoke. This allows authenticated low-privileged users to execute MODIFY operations that should be denied by pyLoad's own permission model. This vulnerability is fixed in 0.5.0b3.dev97.

pyload

marimo is a reactive Python notebook. Prior to 0.23.0, Marimo has a Pre-Auth RCE vulnerability. The terminal WebSocket endpoint /terminal/ws lacks authentication validation, allowing an unauthenticated attacker to obtain a full PTY shell and execute arbitrary system commands. Unlike other WebSocket endpoints (e.g., /ws) that correctly call validate_auth() for authentication, the /terminal/ws endpoint only checks the running mode and platform support before accepting connections, completely skipping authentication verification. This vulnerability is fixed in 0.23.0.

coreweave

cryptography is a package designed to expose cryptographic primitives and recipes to Python developers. From 45.0.0 to before 46.0.7, if a non-contiguous buffer was passed to APIs which accepted Python buffers (e.g. Hash.update()), this could lead to buffer overflows. This vulnerability is fixed in 46.0.7.

cryptography.io

PraisonAI is a multi-agent teams system. Prior to 1.5.115, execute_code() in praisonaiagents.tools.python_tools defaults to sandbox_mode="sandbox", which runs user code in a subprocess wrapped with a restricted __builtins__ dict and an AST-based blocklist. The AST blocklist embedded inside the subprocess wrapper (blocked_attrs of python_tools.py) contains only 11 attribute names — a strict subset of the 30+ names blocked in the direct-execution path. The four attributes that form a frame-traversal chain out of the sandbox are all absent from the subprocess list (__traceback__, tb_frame, f_back, and f_builtins). Chaining these attributes through a caught exception exposes the real Python builtins dict of the subprocess wrapper frame, from which exec can be retrieved and called under a non-blocked variable name — bypassing every remaining security layer. This vulnerability is fixed in 1.5.115.

praison

NiceGUI is a Python-based UI framework. Prior to 3.10.0, Since PurePosixPath only recognizes forward slashes (/) as path separators, an attacker can bypass this sanitization on Windows by using backslashes (\) in the upload filename. Applications that construct file paths using file.name (a pattern demonstrated in NiceGUI's bundled examples) are vulnerable to arbitrary file write on Windows. This vulnerability is fixed in 3.10.0.

zauberzeug · microsoft

rfc3161-client is a Python library implementing the Time-Stamp Protocol (TSP) described in RFC 3161. Prior to 1.0.6, an Authorization Bypass vulnerability in rfc3161-client's signature verification allows any attacker to impersonate a trusted TimeStamping Authority (TSA). By exploiting a logic flaw in how the library extracts the leaf certificate from an unordered PKCS#7 bag of certificates, an attacker can append a spoofed certificate matching the target common_name and Extended Key Usage (EKU) requirements. This tricks the library into verifying these authorization rules against the forged certificate while validating the cryptographic signature against an actual trusted TSA (such as FreeTSA), thereby bypassing the intended TSA authorization pinning entirely. This vulnerability is fixed in 1.0.6.

trailofbits

XWiki Platform is a generic wiki platform offering runtime services for applications built on top of it. Prior to 17.4.8 and 17.10.1, an improperly protected scripting API allows any user with script right to bypass the sandboxing of the Velocity scripting API and execute, e.g., arbitrary Python scripts, allowing full access to the XWiki instance and thereby compromising the confidentiality, integrity and availability of the whole instance. Note that script right already constitutes a high level of access that we don't recommend giving to untrusted users. This vulnerability is fixed in 17.4.8 and 17.10.1.

xwiki

Emmett is a full-stack Python web framework designed with simplicity. From 2.5.0 to before 2.8.1, the RSGI static handler for Emmett's internal assets (/__emmett__ paths) is vulnerable to path traversal attacks. An attacker can use ../ sequences (eg /__emmett__/../rsgi/handlers.py) to read arbitrary files outside the assets directory. This vulnerability is fixed in 2.8.1.

emmett

mise manages dev tools like node, python, cmake, and terraform. From 2026.2.18 through 2026.4.5, mise loads trust-control settings from a local project .mise.toml before the trust check runs. An attacker who can place a malicious .mise.toml in a repository can make that same file appear trusted and then reach dangerous directives such as [env] _.source, templates, hooks, or tasks.

jdx