Cisco-Meraki MCP Server
About
Official Cisco MCP server for connecting AI agents to Meraki's cloud-managed networking solutions.
Details
- Author
- Unknown
- Categories
- Cloud Service, Infrastructure, Other
Jump to
The Meraki MCP server connects AI agents and coding assistants directly to the Meraki Dashboard API, enabling natural language queries and agentic network automation through a standardized interface.
Use it to build agents that answer questions about your network, generate reports, troubleshoot connectivity issues, and audit configurations - without writing custom API integration code.
Beta notice:Meraki MCP is currently in beta. Features, tool schemas, supported API capabilities, and client configuration may change, including in ways that are not backward compatible. We reserve the right to make breaking changes during the beta period in order to incorporate feedback and improvements.
The Meraki MCP server implements this standard for the Meraki Dashboard API. Your AI agent calls MCP tools; the server translates those calls into authenticated Meraki API requests and returns the results.
Meraki MCP is available as a Cisco-hosted remote service or as a self-hosted, open-source server. Both provide the same core MCP functionality; choose the deployment model that best fits your operational and compliance requirements.
Important:You are responsible for any changes you make to the local, open-source MCP. Meraki does not support modified local MCP code.
Your use of the Cisco Meraki MCP Server and any Cisco API leveraged by the MCP Server is governed by theCisco Networking Model Context Protocol (MCP) Server Terms,Cisco Baseline API License Terms,Cisco Acceptable Use Policy,Cisco Meraki Cloud Service Offer Description, andCisco General Terms.
- A Meraki Dashboard account
- A Meraki API key (generate one inDashboard → Organization → API & Webhooks)
- An MCP-compatible client (Claude Desktop, Cursor, VS Code, etc.)
Warning: The Meraki MCP server supports only read operations. However, AI agents are designed to be resourceful and may attempt to call the Meraki Dashboard API directly when they have access to an API key. If the agent is given a read-write key, those direct API calls could make changes outside the MCP server. Give the agent an API key associated with read-only Dashboard access to ensure it cannot make changes.
The Meraki MCP server authenticates using a Meraki Dashboard API key passed as a bearer token.
Authorization: Bearer <MERAKI_DASHBOARD_API_KEY>
Before adding the server in Cursor, VS Code, or Codex, setMERAKI_DASHBOARD_API_KEYin the environment used to launch that client. Restart the client after setting or changing the variable.
export MERAKI_DASHBOARD_API_KEY="your-meraki-dashboard-api-key"
Claude Desktop does not support configuring remote MCP servers throughclaude_desktop_config.json. Add remote servers throughSettings → Connectorsinstead. This flow currently supports unauthenticated and OAuth-based servers; an API-key-authenticated Meraki MCP connection is not yet supported in Claude Desktop.
{ "mcpServers": { "meraki": { "type": "http", "url": "https://mcp.meraki.com/mcp", "headers": { "Authorization": "Bearer ${MERAKI_DASHBOARD_API_KEY}" }, "env": { "MERAKI_DASHBOARD_API_KEY": "${MERAKI_DASHBOARD_API_KEY}" } } } }
Add the following to your.cursor/mcp.json:
{ "mcpServers": { "meraki": { "url": "https://mcp.meraki.com/mcp", "headers": { "Authorization": "Bearer ${env:MERAKI_DASHBOARD_API_KEY}" } } } }
Add the following to your.vscode/mcp.jsonto avoid hardcoding your API key:
{ "servers": { "meraki": { "type": "http", "url": "https://mcp.meraki.com/mcp", "headers": { "Authorization": "Bearer ${env:MERAKI_DASHBOARD_API_KEY}" } } } }
Add the following to your~/.codex/config.toml:
[mcp_servers.meraki] url = "https://mcp.meraki.com/mcp" bearer_token_env_var = "MERAKI_DASHBOARD_API_KEY"
The MCP server exposes curated workflow tools rather than one tool per API endpoint. This keeps the tool surface small enough for AI clients to reason about effectively.
Tip:Be specific in your prompts. The more context you give the agent about what you're looking for, the more precisely it can select and sequence the right tools.
"Generate a switch port stability report across the organization. Find switch ports with link flaps, STP changes, high error rates, or excessive packet loss in the last 7 days. Show the top 10 switches and ports by instability count, identify recurring issues, and recommend what to investigate first."
The agent callssemantic_searchto find switch-event, port-status, and client capabilities, then callsexecute_apifor the selected capabilities to rank unstable ports and identify affected networks or devices.
Wireless Client Experience Degradation Report
"Generate a wireless client experience report for the organization. Identify where clients had poor performance, including low SNR, high retries, roaming, DHCP, DNS, and authentication failures. Show the top 5 networks and access points by impacted client count, highlight failure patterns, and recommend how to improve reliability."
The agent callssemantic_searchto find wireless health, failed-connection, access-point, and SSID capabilities, then callsexecute_apifor the selected capabilities to identify and rank the affected networks and access points.
"Generate an MX WAN health report across the organization. Identify WAN uplink degradation, failovers, packet loss, latency spikes, jitter, and VPN interruptions over the last 30 days. Show the top 5 networks by impact, identify recurring failover or brownout behavior, and recommend how to improve resiliency."
The agent callssemantic_searchto find appliance, uplink, connectivity-event, and VPN capabilities, then callsexecute_apifor the selected capabilities to identify affected sites, circuits, and recurring failure patterns.
Switch Capacity and Oversubscription Report
"Generate a switching capacity report across the organization. Identify switches, uplinks, and ports approaching capacity based on utilization, packet drops, errors, PoE usage, and client density. Highlight congestion or power-exhaustion risk, and recommend where upgrades, rebalancing, or configuration changes are needed."
The agent callssemantic_searchto find switch-port, traffic, power, and client capabilities, then callsexecute_apifor the selected capabilities to identify capacity risk across networks, closets, and devices.
Security Threat and Policy Effectiveness Report
"Generate a security appliance threat and policy report across the organization. Analyze IDS/IPS events, malware and content-filtering blocks, AMP detections, and suspicious outbound traffic over the last 30 days. Show the top 5 networks by security event count, identify recurring threats or policy gaps, and recommend how to reduce risk."
The agent callssemantic_searchto find security-event, appliance, client, and network capabilities, then callsexecute_apifor the selected capabilities to identify impacted networks, clients, and recurring security patterns.
The Meraki MCP server respects theMeraki Dashboard API rate limits:
- Default limit:10 requests per second per organization
- Backoff behavior:The server automatically retries with exponential backoff on429 Too Many Requestsresponses
- Per-org isolation:Rate limits are applied per organization; multi-org sessions do not share a single bucket
This section applies to the Cisco-hosted remote MCP server. If your organization restricts the IP addresses from which Meraki Dashboard API calls can be made, add the following hosted MCP server egress IP addresses to your allowlist:
158.115.141.245 158.115.141.238 158.115.141.209 158.115.133.170 158.115.133.139 158.115.133.156
The local, open-source MCP server makes Dashboard API calls from the IP address of the environment where it is running. Allowlist that address according to your organization's requirements.
Important:The hosted Meraki MCP server does not currently enforce your organization's Dashboard API IP restrictions. Requests made through the hosted MCP server appear to originate from the MCP server's IP address, so a user with a valid API key can make API calls through MCP even when their own IP address is not on the allowlist.
Support for enforcing the same IP restrictions used for direct Dashboard API calls is planned, but is not currently available.
- Read-only:Write operations are not supported in the current release.
- API key auth only:OAuth is not yet supported; bearer token required.
- Hosted MCP: Meraki.com only:Federal, GovCloud, and localized Meraki environments are not supported by the Cisco-hosted MCP service. The local, open-source MCP can work with any Meraki environment.
- IP restrictions:Dashboard API IP restrictions are not enforced for requests made through the hosted MCP server. SeeIP Address Restrictionsfor the required allowlist and local-MCP details.
A comprehensive MCP server for configuring and managing Cisco ACI (Application Centric Infrastructure) fabrics through the APIC REST API.
An MCP server for Cisco NSO (Network Services Orchestrator) that exposes NSO data and operations as MCP primitives.
Access Cisco Support APIs for bug searches and other support-related tasks.
Rent a real Linux VPS as an AI agent — pay USDC (x402 on Base) or card, SSH in under 60s, no signup; provision, exec, firewall, and storage.
An MCP server that enables LLMs to interact with Cisco Intersight APIs.
A Python-based MCP server for Cisco's Meraki Dashboard, providing tools to query the API for discovering, monitoring, and managing your Meraki environment.
An MCP server for Cisco NSO that exposes its data and operations as MCP primitives.
Connect to, configure, and monitor Cisco network devices like routers and switches via SSH.
Production-ready MCP server for AI-driven console automation and monitoring. 40+ tools for session management, SSH, testing, and background jobs.
Hetzner Cloud MCP Server — (Cloud API + SSH)
Hetzner Cloud MCP Server — two management layers (Cloud API + SSH) with 60 tools. Manage server power, snapshots, firewalls, DNS, plus SSH into servers for service control, log viewing, Nginx management, MySQL queries, and system monitoring. Self-hosted PHP, MIT licensed.
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
