# Runlayer MCP for AI Agents AI Agent Connect > Runlayer is an enterprise control plane for managing your Connector ecosystem. It lets you register servers, manage agent skills, and enforce security policies across your whole organization. Use it to govern how your AI agents interact with data, audit every action, and find unauthorized shadow AI tools before they become a risk. ## Overview - **Category:** friends-mcp - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_PaTg4jiaCBAOZI6pcScShh9467XvfcKP4495NMGs/ai-agent-connect - **Tags:** mcp-server, governance, policy-management, agent-lifecycle, security-guardrails, infrastructure-control ## Description Runlayer gives you a way to actually control how your organization uses AI without letting things get messy. If you're trying to manage dozens of Connectors across different teams, you know how hard it is to keep track of who has access to what. This Connector lets you treat your entire AI infrastructure like a production environment. You can tell your agent to onboard new servers, assign specific skills to different teams, and set hard boundaries on what data your agents can touch. It's about moving away from manual dashboard clicking and into a world where you can just give a command to audit your logs or find out which departments are secretly using unauthorized AI tools. Because you're using Vinkius, you can plug this right into your existing workflow to keep your enterprise compliant while your teams actually get work done. It's the control plane you need when just letting everyone use whatever they want isn't an option anymore. You get a single point of truth for your infrastructure, making it easier to rotate keys, update policies, and keep your security posture tight without the headache of manual overhead. ## Tools ### create_mcp_server Onboard a new Connector by providing its name and connection details. You can also assign skills and policies during the registration process. ### create_policy Define new security rules or access requirements for your Connector ecosystem. Use this to restrict access to sensitive data or servers. ### create_skill Register a new reusable capability that your agents can use across different tasks. This helps you codify agent behaviors for consistency. ### delete_agent Remove an AI agent from your organization and disconnect it from all servers. This action clears all associated policy assignments. ### delete_mcp_server Disconnect a specific Connector and remove all associated agent permissions. Use this to clean up old or deprecated endpoints. ### delete_policy Remove a security policy so that resources are no longer governed by its rules. This is useful when updating your security standards. ### delete_skill Delete a specific skill definition without removing the underlying server tools. Use this to prune your list of available agent capabilities. ### get_agent View the configuration and access patterns for a specific registered agent. Use this to troubleshoot connectivity or audit specific users. ### get_audit_logs Pull the full history of actions, identities, and outcomes for your organization. This is essential for compliance and security investigations. ### get_mcp_server Review the configuration and security compliance of a specific Connector. Use this to verify that your endpoints meet company standards. ### get_scan_results View the findings from a shadow AI discovery scan to find policy violations. Use this to see which unauthorized tools were detected. ### get_skill Check the configuration and dependencies for a specific registered skill. This helps you understand how different capabilities are linked. ### list_api_keys See all active keys in your organization to prepare for rotation or audits. This helps you identify unused keys that might be a risk. ### list_agents See every agent in your organization along with their active skills and access. Use this to understand your current agent ecosystem. ### list_mcp_servers View your entire inventory of Connectors and their current status. This is the first step for auditing your infrastructure. ### list_members Audit the members of your organization and their assigned roles. Use this to review who has access to your AI control plane. ### list_policies Review all active security policies and see how many violations have occurred. This gives you a quick snapshot of your security posture. ### list_skills See all registered capabilities and how often they've been used. Use this to discover which skills are most popular across the team. ### revoke_api_key Immediately cancel a compromised or unused API key using its unique ID. This is a critical step for handling security incidents. ### run_mcp_sweep_scan Start a security scan to find unauthorized AI tools across your organization. This proactively identifies shadow AI risks. ### update_mcp_server Change the connection details or policy assignments for an Connector. Use this to rotate credentials or update endpoints. ### update_policy Refine security requirements or update access controls on an existing policy. Use this to tighten or loosen security rules as needed. ### update_skill Adjust the definition or documentation for a specific registered skill. Use this to keep your agent capabilities up to date. ### create_api_key Create a new API key for your Runlayer organization and get the metadata immediately. Use this to set up integrations for your CI/CD pipelines. ### get_organization Get a high-level overview of your entire Runlayer organization and ecosystem. This is a great way to verify your current configuration. ### update_agent Modify the configuration or metadata for an existing registered agent. Use this to update assignments or change agent types. ### create_agent Register a new AI agent with specific types like Cursor or VS Code to the control plane. This helps you onboard agents with proper security guardrails. ## Prompt Examples **Prompt:** ``` What's our current security posture for AI agents? ``` **Response:** ``` ### Current Security Posture Here is the summary of our active policies and recent violations: | Policy Name | Status | Violations | Affected Resources | | :--- | :--- | :--- | :--- | | **Data Privacy Guard** | Active | 2 | SQL_Server_01, HR_Database | | **Dev_Tool_Access** | Active | 0 | VS_Code_MCP | | **External_API_Block** | Active | 5 | Various | **Recent Violations:** - *User_ID_442* attempted to access `HR_Database` via `Cursor_Agent` (Blocked). - *User_ID_109* attempted to call an unapproved external API (Blocked). ``` **Prompt:** ``` Find any unauthorized AI tools in our organization. ``` **Response:** ``` I've completed a **Shadow AI Sweep Scan**. Here are the unauthorized resources discovered: ⚠️ **Unauthorized Connectors Found:** - `test-server-01` (Unregistered Endpoint) - `dev-sandbox-api` (No Policy Assigned) **Risk Level:** High **Recommended Action:** Use `delete_mcp_server` to disconnect these endpoints and update your security policy to block these specific URLs. ``` **Prompt:** ``` How many agents are currently using the 'data_analysis' skill? ``` **Response:** ``` There are currently **14 agents** registered with the `data_analysis` skill. **Breakdown by Client:** - **Claude Desktop:** 6 agents - **Cursor:** 5 agents - **Custom Agents:** 3 agents All agents currently meet the compliance requirements for this skill. ``` ## Capabilities ### Register Connectors Add new serverless endpoints to your organization's control plane. ### Create security policies Set rules for who can access which data and which tools. ### Audit logs View every action your agents take for compliance reporting. ### Scan for shadow AI Find unauthorized AI tools in your network before they cause issues. ### Manage agent skills Define reusable capabilities for your agents to ensure consistency. ### Rotate API keys Keep your organization's credentials secure with automated rotation. ## Use Cases ### Compliance Audit An auditor asks for a report on all agent actions. You use get_audit_logs to pull a clean history of every interaction and identify who performed which action. ### Shadow AI Hunt A security lead suspects unauthorized AI use. They run run_mcp_sweep_scan to identify every rogue Connector on the network and shut them down. ### Onboarding a New Team A platform engineer needs to give a new team access to a tool. They use create_mcp_server and create_policy to set it up safely with specific guardrails. ### Agent Migration A company moves from one client to another. They use list_agents and update_agent to migrate configurations without losing existing settings. ## Benefits - Automated Audit Trails: Use get_audit_logs to get a complete history of every action your agents take for compliance reporting. - Shadow AI Detection: Run run_mcp_sweep_scan to find unauthorized AI tools and prevent data leaks before they happen. - Centralized Policy Management: Use create_policy to set global security standards that apply to every agent in your company. - Skill Standardization: Use create_skill to make sure every agent uses the same high-quality capabilities for consistent results. - Secure Key Rotation: Use list_api_keys and revoke_api_key to manage your credentials without manual overhead. ## How It Works The bottom line is you replace manual dashboard clicking with natural language commands to manage your entire AI infrastructure. 1. Generate an Organization API Key from your Runlayer dashboard. 2. Configure your organization's Runlayer host URL in your AI client. 3. Ask your agent to manage servers, enforce policies, or audit logs. ## Frequently Asked Questions **What is Runlayer MCP for?** Runlayer is an enterprise control plane that lets you manage your entire AI infrastructure. It gives you a central place to register servers, manage agent skills, and enforce security policies across your organization. **Can I use Runlayer MCP to find unauthorized AI tools?** Yes. You can use the sweep scan feature to proactively discover shadow AI and unauthorized Connectors being used across your network. **How does Runlayer MCP help with security compliance?** It provides complete audit logs of every action your agents take. This makes it easy to generate reports for auditors and ensure your AI usage stays within company rules. **Can I manage my API keys through Runlayer MCP?** Yes. You can create, list, and revoke API keys for your organization directly through the control plane to keep your credentials secure. **How does Runlayer MCP handle agent permissions?** You can create specific security policies that define exactly which agents have access to which Connectors and skills, ensuring a zero-trust approach to AI. **Is Runlayer MCP good for large organizations?** It's specifically built for enterprise scale. It's designed for companies managing many servers and agents who need centralized governance and visibility. **Do I need a Runlayer enterprise subscription to use this Connector?** Yes, this Connector requires an active Runlayer organization with API access. Runlayer is an enterprise-grade control plane, so you need a valid organizational subscription. Contact Runlayer sales to get started and obtain your organization API key. **Can this Connector detect unauthorized AI usage (shadow AI)?** Yes! The run_mcp_sweep_scan tool initiates comprehensive shadow AI discovery across devices, detecting unauthorized Connectors, OpenClaw installs, Skills, and agents. Results include policy violations and security risks across your organization's endpoints. **What types of AI agents does Runlayer support?** Runlayer supports Claude Desktop, Cursor, VS Code with Copilot, Windsurf, and custom AI agents. Each agent type can be registered with specific security policies, assigned Connectors, and monitored through the audit trail. New agent types can be added as custom integrations. **Do I need a Runlayer enterprise subscription to use this MCP?** Yes, this MCP server requires an active Runlayer organization with API access. Runlayer is an enterprise-grade control plane, so you need a valid organizational subscription. Contact Runlayer sales to get started and obtain your organization API key. **Can this MCP server detect unauthorized AI usage (shadow AI)?** Yes! The run_mcp_sweep_scan tool initiates comprehensive shadow AI discovery across devices, detecting unauthorized MCP servers, OpenClaw installs, Skills, and agents. Results include policy violations and security risks across your organization's endpoints.