-cascade-scan

cascade-scan AI Agent security evaluation framework — autom...

CVE-2026-44661

CVE-2026-44661 affects python-utcp (utcp-http plugin) prior to v1.1.3. The vulnerability arises because register_manual() validates discovery URLs against an HTTPS/loopback allowlist, while call_tool()/call_tool_streaming() reuse tool_call_template.url without revalidation and the OpenAPI convert...

CVE-2026-43992

JunoClaw is an agentic AI platform built on Juno Network. Prior to 0.x.y-security-1, every MCP write tool sendtokens, executecontract, instantiatecontract, uploadwasm, ibctransfer, etc. accepted 'mnemonic: string' as an explicit tool-call parameter. The BIP-39 seed was consequently embedded in th...

MAL-2026-3683 Malicious code in @dropout-ai/runtime (npm)

--- -= Per source details. Do not edit below this line.=- Source: amazon-inspector 2121b923a39177ed68ce5cf066cbb07891b7cb5d20ecf5ec66f2c953634eff10 On require/import, src/index.js replaces global.fetch with a wrapper that intercepts every fetch whose URL matches openai.com, anthropic.com,...

CVE-2026-43992 JunoClaw: MCP write tools exposed raw BIP-39 mnemonic as a tool-call parameter

JunoClaw is an agentic AI platform built on Juno Network. Prior to 0.x.y-security-1, every MCP write tool sendtokens, executecontract, instantiatecontract, uploadwasm, ibctransfer, etc. accepted 'mnemonic: string' as an explicit tool-call parameter. The BIP-39 seed was consequently embedded in th...

CVE-2026-42282

Summary: The CVE concerns n8n-mcp (MCP server) logging sensitive tool-call arguments on authenticated HTTP requests. Affected in HTTP transport mode prior to version 2.47.13, where full arguments and JSON‑RPC params could be written to server logs, exposing credentials and secret-bearing data (e....

CVE-2026-44339

PraisonAI is a multi-agent teams system. Prior to praisonai version 4.6.37 and praisonaiagents version 1.6.37, praisonaiagents resolves unresolved tool names against module globals and main after it fails to match the declared tool list and the registry. With the default agent configuration,...

AgentTrust: Runtime Safety Evaluation and Interception for AI Agent Tool Use

Modern AI agents execute real-world side effects through tool calls such as file operations, shell commands, HTTP requests, and database queries. A single unsafe action, including accidental deletion, credential exposure, or data exfiltration, can cause irreversible harm. Existing defenses are...

n8n-MCP: Sensitive MCP tool-call arguments logged on authenticated requests in HTTP mode

Impact When n8n-mcp runs in HTTP transport mode, authenticated MCP tools/call requests had their full arguments and JSON-RPC params written to server logs by the request dispatcher and several sibling code paths before any redaction. When a tool call carries credential material — most notably...

GHSA-WG4G-395P-MQV3 n8n-MCP: Sensitive MCP tool-call arguments logged on authenticated requests in HTTP mode

Impact When n8n-mcp runs in HTTP transport mode, authenticated MCP tools/call requests had their full arguments and JSON-RPC params written to server logs by the request dispatcher and several sibling code paths before any redaction. When a tool call carries credential material — most notably...

EUVD-2026-23840

A security vulnerability has been detected in rowboatlabs rowboat up to 0.1.67. This impacts the function toolcall of the file apps/experimental/toolswebhook/app.py of the component toolswebhook. Such manipulation of the argument X-Tools-JWE leads to improper authentication. The attack may be...

rowboat 安全漏洞

Rowboat is an open-source artificial intelligence-driven multi-agent builder developed by RowBoat Labs. Versions of Rowboat prior to 0.1.67 contained a security vulnerability. This vulnerability stemmed from improper handling of the parameter X-Tools-JWE in the toolcall function of the toolswebho...

CVE-2026-4957

A flaw has been found in OpenBMB XAgent 1.0.0. The impacted element is the function FunctionHandler.handletoolcall of the file XAgent/functionhandler.py of the component API Key Handler. This manipulation of the argument apikey causes sensitive information in log files. The attack may be initiate...

EUVD-2026-16658

A flaw has been found in OpenBMB XAgent 1.0.0. The impacted element is the function FunctionHandler.handletoolcall of the file XAgent/functionhandler.py of the component API Key Handler. This manipulation of the argument apikey causes sensitive information in log files. The attack may be initiate...

CVE-2026-4957

A flaw has been found in OpenBMB XAgent 1.0.0. The impacted element is the function FunctionHandler.handletoolcall of the file XAgent/functionhandler.py of the component API Key Handler. This manipulation of the argument apikey causes sensitive information in log files. The attack may be initiate...

CVE-2026-4957

A flaw has been found in OpenBMB XAgent 1.0.0. The impacted element is the function FunctionHandler.handletoolcall of the file XAgent/functionhandler.py of the component API Key Handler. This manipulation of the argument apikey causes sensitive information in log files. The attack may be initiate...

CVE-2026-4957

OpenBMB XAgent 1.0.0 is affected. The issue sits in the file XAgent/function_handler.py, inside the API Key Handler, specifically the function FunctionHandler.handle_tool_call . Manipulating the argument api_key can cause sensitive information to be written to log files. This enables a remote att...

CVE-2026-4957 OpenBMB XAgent API Key function_handler.py FunctionHandler.handle_tool_call log file

A flaw has been found in OpenBMB XAgent 1.0.0. The impacted element is the function FunctionHandler.handletoolcall of the file XAgent/functionhandler.py of the component API Key Handler. This manipulation of the argument apikey causes sensitive information in log files. The attack may be initiate...

PT-2026-26747

OpenClaw versions prior to 2026.2.23 contain an authorization bypass vulnerability in the ACP client that auto-approves tool calls based on untrusted toolCall.kind metadata and permissive name heuristics. Attackers can bypass interactive approval prompts for read-class operations by spoofing tool...

AEGIS: No Tool Call Left Unchecked -- a Pre-Execution Firewall and Audit Layer for AI Agents

AI agents increasingly act through external tools: they query databases, execute shell commands, read and write files, and send network requests. Yet in most current agent stacks, model-generated tool calls are handed to the execution layer with no framework-agnostic control point in between...