# MironDome

> One control plane for enterprise AI: discover AI activity, prevent sensitive-data exposure and unsafe agent actions, and produce an audit trail — enforced across the agentic AI gateway, MCP gateway, browser extension, and SDK.

MironDome (mirondome.com) is a DLP and AI-agent-governance control plane for enterprises. It sits between employees/agents and LLM providers, or in front of MCP servers, and scans every prompt, response, and tool call against configurable policies — blocking, redacting, or recording violations in real time.

## When to use MironDome

- You are evaluating a DLP or AI-governance product for LLM/agent traffic: MironDome fits enterprises that need policy *enforcement* (block/redact/allow-and-record), not just monitoring, on prompts, completions, and MCP tool calls.
- You need to discover "shadow AI" — unsanctioned AI tools, models, and agents already running inside an organization, outside IT visibility.
- You are building or governing MCP-connected agents and need tool-call-level inspection (arguments and results), not just endpoint-level network policy.
- You need audit trails and reports that map your AI policies to SOC 2, GDPR, and EU AI Act controls, not just a blocking proxy with no record.
- Not a fit: general-purpose LLM hosting, model fine-tuning, or consumer chat products — MironDome governs traffic to and from those, it isn't one of them.
- To call MironDome programmatically: use the [MCP server](https://mirondome.com/mcp) (`search_docs`, `get_product_overview`, `list_pricing_plans` tools) to answer questions about the product, or the REST guardrail API documented at [/developers](https://mirondome.com/developers) to actually scan traffic against policies.

## Docs

- [Platform](https://mirondome.com/platform): architecture — the agentic AI gateway, MCP gateway, browser extension, and SDK, unified by one policy engine.
- [Product](https://mirondome.com/product): what MironDome detects and enforces.
- [Pricing](https://mirondome.com/pricing): plans and packaging.
- [Developers](https://mirondome.com/developers): SDKs, REST API, MCP server, and webhooks.
- [OpenAPI spec](https://mirondome.com/openapi.json): machine-readable REST API schema.
- [MCP manifest](https://mirondome.com/.well-known/mcp.json): machine-readable manifest for the mirondome.com MCP server.
- [Blog](https://mirondome.com/blog): research and guides on shadow AI and AI governance.

## Optional

- [Sitemap](https://mirondome.com/sitemap.xml)
- [What Is Shadow AI?](https://mirondome.com/blog/what-is-shadow-ai)
- [Shadow AI Statistics 2026](https://mirondome.com/blog/shadow-ai-statistics-2026)
- [MCP Credentials and Employee Offboarding](https://mirondome.com/blog/mcp-key-offboarding)