Lasso introduces first security gateway for MCP workflows

Lasso has launched the first security gateway designed for Model Context Protocol (MCP), focusing on the security needs of generative AI workflows.

The Lasso MCP Gateway operates as a proxy and orchestrator for MCP interactions, embedding essential security, governance, and monitoring measures while supporting compatibility with existing agent implementations.

Since its introduction by Anthropic in November 2024, MCP has become a widely adopted standard for linking generative AI models to external tools and data sources. Thousands of developers have contributed to the MCP ecosystem, with the Awesome MCP Servers repository on GitHub gathering more than 37,000 stars, underscoring significant community engagement.

Despite rapid adoption, security and visibility vulnerabilities remain prevalent in agentic workflows. Lasso aims to address these vulnerabilities by introducing foundational security tools at a time of substantial growth in the GenAI environment.

"When MCP emerged, we immediately saw its revolutionary potential, but also recognized its security blind spots," said Lior Ziv, CTO and Co-founder, Lasso. "That is why we built the world's first MCP gateway with security built-in to enable the continued innovation of agentic workflows. Lasso MCP Gateway makes it possible to scale GenAI agent development without compromising on safety, confidentiality, and transparency."

The Lasso MCP Gateway's architecture includes several key features. Its agnostic guardrails allow for the application of configurable security filters to both requests and responses, reducing the risk of sensitive data exposure before information reaches agents. The tool works consistently with any connected MCP regardless of inherent security capabilities.

The platform also offers advanced tracking functionalities, delivering usage analytics and pattern identification to help developers optimise their environments. A unified visibility dashboard enables users to monitor all MCPs from a single interface for comprehensive oversight.

Enterprise readiness is another aspect of the MCP Gateway. The product is developed in collaboration with the wider community, allowing organisations to introduce essential enterprise features to MCP deployments as they adopt the protocol for mission-critical applications.

Several risks associated with the expansion of agentic workflows prompted the implementation of these controls. These risks include sensitive data exposure, where outputs that are not properly sanitised may reveal secrets, credentials, or internal model states. Prompt injection is another concern, in which hidden instructions embedded in tool descriptions or parameter hints are inadvertently executed by the model during runtime.

Command injection and remote code execution pose further threats if unvalidated inputs reach shell commands or system calls, potentially enabling immediate remote code execution. In addition, a lack of logging and monitoring creates challenges in tracking tool usage and model-tool interactions, which hinders the detection or investigation of abuse or misuse.

Lasso states that its MCP Gateway directly addresses these vulnerabilities by enabling teams to define policies, monitor risks in real-time, intercept attacks before they escalate, and support the open source community's drive for enterprise-grade security in the adoption of MCP.

The Lasso MCP Gateway is made available as part of the company's GenAI security platform and can be accessed on GitHub and other prominent open source platforms.