MintMCP

MintMCP is a managed SaaS platform that provides governance, access control, and security policy enforcement for AI agent tool use via the Model Context Protocol (MCP). It is SOC 2 Type II certified and operates in both US and EU regions, with VPC and self-hosted deployment available on request. Core capabilities: - Agent Bundles: A single governance object that combines SCIM group membership, tool access scoping, security policy, and audit logging per agent or team. Each agent receives its own OAuth credentials, rotatable independently of user accounts. - Policy enforcement on tool calls: Supports custom policy code executed in a sandboxed runtime on every tool call, with the ability to inspect, transform, mask, or block requests. - Prebuilt DLP integrations: Out-of-the-box integrations with AWS Bedrock Guardrails, GCP DLP, Microsoft Purview, Nightfall, and Skyflow for data loss prevention on tool calls. - Threat detection presets: Built-in rules for secrets detection, prompt injection, and risky bash commands, as well as OpenAI moderation and jailbreak detection templates. - MCP connector catalog: A catalog of 10,000+ MCP servers with auto-scaling, sandboxed, isolated execution per connector. - Shadow AI discovery: An Agent Monitor component that detects off-gateway MCP use in Cursor and Claude Code, supports MDM-pushed enforcement on user devices, and performs DLP inspection of local agent activity including bash, file reads/writes, and prompts. - SSO and SCIM: Okta, Entra, and Google SSO included on all plans; SCIM-driven group mapping for team membership. - Hosted Agents platform: Supports per-agent identity with long-term memory and Slack integration.