Perceptive Security SOC/SIEM Consultancy
🇳🇱 🇬🇧
← Terug naar overzicht
Criticality: Critical · ·Source: nvd.nist.gov

Chainlit is a Python framework for building production-ready conversational AI applications. From 2.4.0rc0 until 2.12.0, Chainlit deployments with features.mcp.enabled set to true in .chainlit/config.toml expose the POST /mcp endpoint without requiring authentication. For stdio transport, the endpoint accepts a user-controlled fullCommand string. The validate_mcp_command() function in backend/chainlit/mcp.py checks only the executable name against config.features.mcp.stdio.allowed_executables and passes unchecked arguments to StdioServerParameters in backend/chainlit/server.py. Because npx supports the -c argument, an attacker can execute arbitrary shell commands with the privileges of the Chainlit process. If allowed_executables is unset, its None default is treated as allowing every executable. This issue is fixed in version 2.12.0.

CVE-2026-45018 affects Chainlit, a Python framework for conversational AI applications, in versions 2.4.0rc0 through 2.12.0. When the MCP feature is enabled, the POST /mcp endpoint is exposed without authentication. The validate_mcp_command() function only checks the executable name but fails to validate arguments passed to StdioServerParameters. An attacker can exploit npx's -c argument to execute arbitrary shell commands with Chainlit process privileges. If allowed_executables is unset, the None default permits any executable, worsening the exposure. The vulnerability requires no authentication to exploit, making it accessible to unauthenticated remote attackers. The issue has been patched in version 2.12.0. Organizations running Chainlit with MCP enabled should upgrade immediately.

Affected products

  • Chainlit 2.4.0rc0 - 2.12.0

Related CVE's

  • CVE-2026-45018

Categories

  • Emerging Technologies
  • Identity & Access
  • Web Technologies

Related links