Agent Sentinel

Infisical Agent Sentinel enables organizations to securely connect AI agents and assistants to external tools through the Model Context Protocol (MCP). It provides a centralized control plane for managing which tools AI systems can access, how they authenticate, and complete visibility into every tool invocation.

As AI agents become more capable and autonomous, organizations need robust infrastructure to govern their access to external systems. Agent Sentinel solves this by acting as a secure gateway between AI clients (like Claude, ChatGPT, or custom agents) and the MCP servers that provide tools.

Core Capabilities

<CardGroup cols={2}> <Card title="MCP Servers" icon="server" href="/documentation/platform/agent-sentinel/mcp-servers"> Connect to remote MCP servers like Notion, GitHub, and Slack with flexible credential management. </Card> <Card title="MCP Endpoints" icon="link" href="/documentation/platform/agent-sentinel/mcp-endpoints"> Create secure connection URLs for AI clients with granular tool selection. </Card> <Card title="Activity Logs" icon="list" href="/documentation/platform/agent-sentinel/activity-logs"> Monitor and audit every tool invocation with detailed request/response logging. </Card> <Card title="Tool Governance" icon="shield-check"> Fine-grained control over which tools are available through each endpoint. </Card> <Card title="PII Filtering" icon="user-shield" href="/documentation/platform/agent-sentinel/mcp-endpoints#pii-filtering"> Automatically redact sensitive data like emails, phone numbers, and credit cards from AI tool interactions. </Card> </CardGroup>