# Security Policy
Connecting AI agents to external tools via MCP servers can introduce security risks if those agents process text input from malicious users. Malicious inputs may exploit prompt injection vulnerabilities, potentially causing the agent to execute unauthorized actions on connected systems.
When using the ChartMogul MCP server, your data may be exposed to large language models (LLMs) as part of processing. Depending on the specific design or vulnerabilities of other MCP servers, data could be further exposed to third-party systems or platforms. Use of all such functionality is at your own risk.
You are responsible for evaluating and implementing appropriate safeguards, including input sanitization, output monitoring, and access controls, to mitigate potential misuse.
