Haldir is an MCP (Model Context Protocol) server designed as a governance, security, and audit layer for AI agents and tools. It exposes a suite of MCP-native capabilities including agent identity/session management, secrets vault, authorization of payments/budgets, immutable audit trails, human-in-the-loop approvals, and policy enforcement via a proxy that intercepts and governs all agent tool calls. Haldir integrates with any MCP-compatible client (such as Claude, Cursor, and Windsurf) and is intended for organizations and teams adopting AI agents where oversight, compliance, secrets management, and spend control are required.
Visit Haldir's official website for product details and getting started.