Loading…
// weekly digest
// weekly digest
2026-07-06 00:00 UTC 2026-07-12 23:59 UTC
// total
1,493
// critical
136
// high
569
// medium
582
// low
141
// new kev
6
// top critical
Ruflo is an agent meta-harness for Claude Code and Codex. Prior to 3.16.3, ruflo's default docker-compose deployment exposed the MCP bridge POST /mcp and POST /mcp/:group endpoints without authentication, allowing an unauthenticated network attacker to invoke tools/call to terminal_execute, obtain a shell in the bridge container, read provider API keys, and poison AgentDB learning-store patterns. This issue is fixed in version 3.16.3.
PraisonAI before 1.6.78 contains a remote code execution vulnerability in CodeAgent._execute_python() that executes LLM-generated Python code without AST validation, import restrictions, or sandbox enforcement. Attackers can influence LLM output through prompt injection to exfiltrate all environment secrets and execute arbitrary code on the host system.
The Joomla extension RSFiles is vulnerable to an unauthenticated arbitrary file upload that allows uploading executable files and leads to full RCE.
Langroid is a framework for building large-language-model-powered applications. Versions prior to 0.65.2 are vulnerable to a critical Sandbox Escape leading to Remote Code Execution (RCE) in its `TableChatAgent` and `VectorStore` capabilities. When these agents evaluate LLM-generated tool messages with `full_eval=True`, they attempt to sandbox the execution by explicitly setting `locals` to an empty dictionary `{}` inside Python's `eval()` function. However, this relies on an incomplete understanding of Python's execution model. Because `__builtins__` is not explicitly scrubbed from the `globals` dictionary mapping, Python implicitly injects all built-ins during execution, granting full access to functions like `__import__('os').system()`. Since `TableChatAgent.pandas_eval()` executes external LLM outputs natively, this bypass permits any attacker providing prompt payload to achieve unauthenticated RCE on the host system. Version 0.65.2 patches the issue.
Balbooa Forms contains an unrestricted upload of file with dangerous type vulnerability that allows an unauthenticated arbitrary file upload which could allow uploading of executable files leading to full RCE.
// top high
Composio SDK before 0.2.32-beta.283 contains a path validation bypass vulnerability that allows attackers to read and exfiltrate sensitive files by exploiting a missing assertSafeFileUploadPath check in the readFileFromDisk function within tool-file-uploads.ts. Attackers can exploit prompt injection to manipulate file_uploadable parameters to reference sensitive paths such as SSH private keys, causing the CLI to upload credential files to attacker-controlled storage.
FOSSBilling is a free, open-source billing and client management system. Versions 0.6.10 through 0.7.2 have a PHP code injection vulnerability in FOSSBilling's `Config::prettyPrintArrayToPHP()` method. When configuration values are updated, string values are written into `config.php` without escaping single quotes. Because `config.php` is loaded via a bare `include` on every HTTP request, an attacker with admin privileges can inject arbitrary PHP code that executes on every subsequent request. Version 0.8.0 contains a patch. Some workarounds are available. Restrict admin access to trusted personnel only; audit `config.php` for unexpected PHP code; and/ or at the reverse proxy/WAF level, restrict access to admin API endpoints that modify configuration.
// new kev additions
Balbooa Forms contains an unrestricted upload of file with dangerous type vulnerability that allows an unauthenticated arbitrary file upload which could allow uploading of executable files leading to full RCE.
iCagenda contains an unrestricted upload of file with dangerous type vulnerability that allows the upload of arbitrary files in the file attachment feature, ultimately resulting in PHP code upload and execution.
A vulnerability was determined in Comfast CF-WR631AX V3 up to 2.7.0.8. Affected by this vulnerability is the function system_wl_upload_pic_file of the file /usr/bin/webmgnt of the component FastCGI Backend. This manipulation of the argument filename causes os command injection. It is possible to initiate the attack remotely. The exploit has been publicly disclosed and may be utilized. The vendor was contacted early about this disclosure but did not respond in any way.
During bosh create-env and bosh delete-env, the CLI uploads compiled CPI packages and rendered job templates to the new VM's DAV blobstore over HTTPS without verifying the server certificate, even though a CA certificate for that endpoint is available in the installation manifest. A network attacker can terminate the TLS connection, harvest the Basic-auth credentials, and read the rendered-templates archive containing every bootstrap secret for the new BOSH Director, then replay the credentials against the real VM's agent for root code execution. Affected versions: bosh-cli versions prior to v7.10.4.
PraisonAI before 1.7.3 contains an insecure default configuration that binds to all interfaces with no API key requirement and wildcard CORS. Unauthenticated attackers can call GET /api/agents to read agent instructions and system prompts, or POST /api/chat to invoke agents without authentication.
Joomlack Page Builder contains an improper access control vulnerability that could allow for remote code execution via unauthenticated arbitrary file upload.
Adobe ColdFusion contains a path traversal vulnerability that could lead to arbitrary code execution in the context of the current user.
JoomShaper SP Page Builder contains an unrestricted upload of file with dangerous type vulnerability that allows unauthenticated users to upload arbitrary files, ultimately resulting in the upload and execution of PHP code.
Langflow contains an authorization bypass through user-controlled key vulnerability which allows an authenticated attacker to execute any flow belonging to another user by specifying the victim's flow ID in the request.