Cisco Meraki + AI - for Claude, the ChatGPT desktop app, GitHub Copilot, Gemini CLI, and any agent that speaks MCP
Unofficial. Community-built Claude Code Skill and MCP server for the Cisco Meraki
API. Not affiliated with, endorsed by, or sponsored by Cisco Systems, Inc.
▶ Watch the 30-second demo - demo data is simulated; every command shown exists in the real CLI.
Every Meraki Dashboard operation, plus cross-org fleet views that read every page, and a local mirror built for MSPs running many client orgs. Works with the AI you already use - ChatGPT (desktop app), Claude Desktop, Codex, Claude Code, Claude Cowork, and GitHub Copilot - plus Gemini CLI. Free, open source, runs on your laptop. Built for MSP owners. No code required.
Works with your agent
The six agents MSP owners actually use (self-serve, works today):
| Your AI agent | How to install the Cisco Meraki skill |
|---|
| Claude Desktop | Run installer, then Settings > Extensions to register cisco-meraki-mcp (no JSON editing). |
| ChatGPT (desktop app) | Run installer, then add cisco-meraki-mcp in Settings > MCP servers as a local (STDIO) server. ChatGPT on the web is not supported yet. |
| Codex CLI | Paste the install prompt below. |
| Claude Code | Paste the install prompt below. |
| Claude Cowork | Paste the install prompt below. |
| GitHub Copilot (VS Code) | Run installer, add cisco-meraki-mcp to mcp.json under the servers key, then pick Agent mode. |
The ChatGPT desktop app runs cisco-meraki-mcp locally and shares MCP configuration with Codex (OpenAI). ChatGPT on the web needs a remote server, which msp-skills does not support yet. If your team works there, a hosted gateway is the better pick; check that your AI app is on the gateway's setup list.
Also for the Microsoft and Google stacks
Big install base, but an honest heads-up: Gemini CLI runs the local binary you just installed, while Microsoft 365 Copilot and the Gemini app need a remote server, which msp-skills does not support yet.
| Agent | What it takes |
|---|
| Microsoft 365 Copilot / Copilot Studio | Not supported by msp-skills yet: it needs a remote server. A hosted gateway is the better pick; check that your AI app is on the gateway's setup list. |
| Google Gemini | Gemini CLI is local - same as Claude Code. See mcp-install.md. The Gemini app needs a remote server and is not supported by msp-skills yet. |
Skill-native agents (also covered): Hermes and OpenClaw read this skill's SKILL.md directly and speak MCP - see their install sections below. Also works with Cursor, Windsurf, Cline, Continue.dev, and Zed via MCP. Full per-tool wire-up: docs/which-agent.md.
Run more than one agent? Install across all 51+ supported agents in one command: npx skills add Servosity/msp-skills@latest (requires Node.js, then run the per-skill installer for the CLI/MCP binaries). See docs/which-agent.md.
Install in 60 seconds
Fastest for Claude Desktop - one-click .mcpb
Download Cisco Meraki MCP (.mcpb) - then open Claude Desktop > Settings > Extensions and select the file. One click, no JSON, no shell. (Browse every Cisco Meraki release on the releases page.)
Prefer the Claude Code plugin? Add the marketplace once, then install - works immediately, no directory listing required:
/plugin marketplace add Servosity/msp-skills
/plugin install cisco-meraki@msp-skills
Path A - paste one prompt into your AI agent (recommended)
Copy this into Claude Code, Codex CLI, or Claude Cowork:
Install the Cisco Meraki Skill and MCP server from Servosity/msp-skills in this agent workspace. If this workspace uses a POSIX shell (macOS, Linux, WSL, or Bash), run bash <(curl -fsSL https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/cisco-meraki/install.sh). If it uses Windows PowerShell, run iwr -useb https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/cisco-meraki/install.ps1 | iex. Then authenticate per the README and run cisco-meraki-cli --help to explore.
The same prompt works in any agent that can run shell.
Path B - run the installer yourself
Windows (PowerShell):
iwr -useb https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/cisco-meraki/install.ps1 | iex
macOS / Linux:
bash <(curl -fsSL https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/cisco-meraki/install.sh)
The installer drops both cisco-meraki-cli (the CLI) and cisco-meraki-mcp (the MCP server) into your user bin path. Claude Code, Codex, and Cowork discover the Skill via SKILL.md in this directory.
Verify:
cisco-meraki-cli --version
Upgrade to the latest version
The installer always fetches the current release - re-run it to upgrade:
macOS / Linux:
bash <(curl -fsSL https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/cisco-meraki/install.sh)
Windows (PowerShell):
iwr -useb https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/cisco-meraki/install.ps1 | iex
Claude Desktop .mcpb users: download the latest .mcpb (top of this section) and re-select it in Settings > Extensions. Claude Code plugin users: /plugin update cisco-meraki@msp-skills.
Add to Claude Desktop, the ChatGPT desktop app, GitHub Copilot, Gemini CLI, or another MCP client
After the installer runs, see mcp-install.md and docs/which-agent.md for the per-agent wire-up - one section per agent, including the GitHub Copilot servers key and the ChatGPT desktop app. Claude Desktop's Settings > Extensions panel is the simplest path; the MCP config block (for users who prefer editing JSON) is documented in mcp-install.md.
Install for Hermes
From the Hermes CLI:
hermes skills install servosity/msp-skills/skills/cisco-meraki --force
Inside a Hermes chat session:
/skills install servosity/msp-skills/skills/cisco-meraki --force
Hermes speaks MCP natively, so it can also use the cisco-meraki-mcp server directly - same install path, same env vars.
Install for OpenClaw
Tell your OpenClaw agent (copy this):
Install the cisco-meraki skill from https://github.com/servosity/msp-skills/tree/main/skills/cisco-meraki. The skill defines how its required CLI (cisco-meraki-cli) can be installed via the openclaw: frontmatter block.
OpenClaw isn't generally available yet; the frontmatter wiring is pre-shipped and will activate the moment OpenClaw launches.
Authenticate
Generate a Dashboard API key (Meraki Dashboard > My profile > API access) and set it:
export MERAKI_DASHBOARD_API_KEY=<your-api-key>
cisco-meraki-cli doctor
cisco-meraki-cli fleet health --limit 5
The key reaches every organization the admin who generated it can see, with that admin's permissions - generate it from a read-only admin if you only want reporting. Each org needs Dashboard API access turned on (Organization > Settings). doctor checks config and reachability and test-calls the key; read its Credentials line, because it exits 0 even when the key is rejected. On a regional dashboard, also set CISCO_MERAKI_BASE_URL (for example https://api.meraki.ca/api/v1 for Canada).
What this skill does
| Question your MSP keeps asking | Command |
|---|
| Which client org is in the worst shape right now? | cisco-meraki-cli fleet health --agent |
| What is offline or alerting across every org, and for how long? | cisco-meraki-cli fleet down --agent |
| Which MX WAN uplinks have failed? | cisco-meraki-cli fleet down --kind uplink --json |
| Which licenses, subscriptions, and end-of-support dates are due in 90 days? | cisco-meraki-cli fleet renewals --within 90d --agent |
| Who changed what across all orgs this week? | cisco-meraki-cli fleet changes --since 7d --group-by admin --agent |
| Did a change land right before this device went offline? | cisco-meraki-cli fleet blame "Q2QN-9J8L-SLPD" --window 2h --agent |
| Which org and network owns this MAC, IP, or serial? | cisco-meraki-cli fleet find "00:18:0a:12:34:56" --agent |
| Which admins have two-factor off or stale access, and where does a departed tech still have access? | cisco-meraki-cli fleet admins --stale 90d --flagged --agent |
Beyond the fleet views, all 1,035 Dashboard API operations are typed commands under organizations, networks, devices, and administered; cisco-meraki-cli which "offline devices" finds the right one.
Full command reference: guide.md. For the AI-agent operating contract (--agent, --dry-run, when to confirm before mutating), see AGENTS.md.
What makes this different
Most Meraki integrations and MCP servers proxy one question into one live API call against one organization, and often read only the first page of a list. That's fine for one record. It dies at MSP scale, when you're asking "what is down for all 30 clients, and since when?"
This skill is built for that question. The fleet commands run across every org the key can see under one rate limiter, follow every Link-header page, and never count an org that fails to answer as zero devices: fleet health keeps it as an unreachable row and the other fleet commands name it on stderr. Compound views do the joins a stateless wrapper can't: fleet down joins device status to org and network names and computes downtime, fleet blame lines up each outage with the config changes made just before it, and fleet renewals merges three licensing models with hardware end-of-support dates into one date-sorted list. The fleet views query the Dashboard live; sync also keeps a local SQLite mirror of the org-level data so search "Acme" --data-source local, fleet find, and fleet changes --data-source local can answer offline. Single-endpoint reads through the MCP server say when Meraki has more pages (has_more, plus the cursor to resume, or how to re-request a page too large for one result).
The pain this closes
Meraki's MSP Portal (now the Global Overview page), as Meraki documents it, lists each client org with device counts, license status, and one co-term expiration date. It documents no cross-org list of which devices are down, no cross-org change log, and no admin management across orgs - so "what is broken for every client right now?" still means clicking into each org. Scripting it runs into the API's shape: paginated lists come back as bare arrays paged only by a Link header, so a one-call script silently stops at page one (the Meraki Community Developers & APIs board has threads like "Using Powershell how to paginate thru pages" on exactly this), all under a ten-calls-per-second, per-org rate limit.
cisco-meraki-cli fleet down --agent - the morning triage list: every offline or alerting device and failed uplink, with network names and downtime.
cisco-meraki-cli fleet renewals --within 90d --agent - the renewal pipeline across co-term, per-device, and subscription orgs plus end-of-support hardware.
cisco-meraki-cli fleet blame "Q2QN-9J8L-SLPD" --window 2h --agent - "did a tech break it?" answered with the changes made just before each outage.
cisco-meraki-cli fleet admins --email "departed.tech@msp.example" --json - every org that still grants a departed tech access.
See pain-point.md for the longer narrative.
Frequently asked questions
Does this work with ChatGPT?
Yes, in the ChatGPT desktop app. It runs MCP servers locally (STDIO), the way cisco-meraki-mcp runs, and shares MCP configuration with Codex (OpenAI): add cisco-meraki-mcp under Settings > MCP servers as a STDIO server, or add it once to Codex's ~/.codex/config.toml. ChatGPT on the web is not supported by msp-skills yet: it needs a remote server. If your team works there, a hosted gateway is the better pick; check that your AI app is on the gateway's setup list.
Does this work with Codex, Cursor, Windsurf, Cline, Copilot, or Gemini?
Yes - all of them speak MCP. Cross-tool install commands are in the matrix above and the deep-dive in docs/which-agent.md.
Do I need to know how to code?
No. The recommended install is to paste one sentence into Claude Code or Codex - your agent reads SKILL.md and does the install. The fallback is a one-line installer per OS (bash or PowerShell). Neither path requires writing code. You'll enter your Cisco Meraki credentials once.
Is my Cisco Meraki data safe?
Your data stays on your machine. The CLI and MCP server are local binaries. The SQLite mirror sits in a directory under your user account. The AI agent only sees what the CLI returns - typically a query result, not raw bulk data. Credentials are read from your environment or your agent's config; never bundled into this repo or transmitted anywhere by MSP Skills.
What can the API key reach?
A Meraki API key carries the Dashboard permissions of the admin account that generated it, across every organization that admin can see. Generate it from an admin with read-only access if you only want reporting. Each org must also have Dashboard API access enabled (Organization > Settings); an org without it shows up in fleet health as reachable: false rather than disappearing.
Will this hit my Meraki API rate limit?
Meraki allows about ten calls per second per organization. The CLI paces itself at 10 requests per second by default (change it with --rate-limit) and, when Meraki answers HTTP 429, waits for the Retry-After header before trying again. Fleet commands across many orgs take longer: each run has one overall --timeout (default 1m), so raise it (for example --timeout 10m) for a large fleet. An org that runs out of time is named on stderr as failed, not silently dropped.
Does it work with the Canada, China, India, or FedRAMP dashboards?
Yes. Set CISCO_MERAKI_BASE_URL to your region's API host: https://api.meraki.ca/api/v1 (Canada), https://api.meraki.cn/api/v1 (China), https://api.meraki.in/api/v1 (India), or https://api.gov-meraki.com/api/v1 (FedRAMP). Leave it unset for the global dashboard.
Can an agent change my clients' networks by accident?
From the CLI, writes send immediately when called, with no confirmation prompt. Over MCP, every write tool returns a preview and sends nothing until it is called again with confirm set to boolean true. Keep autonomous agents to reads and --dry-run previews. The bulk tool, fleet push, previews a per-network diff by default and sends nothing unless you add --apply. Set CISCO_MERAKI_MCP_READ_ONLY=1 on the MCP server and it refuses writes outright. The safety model below says which commands to keep behind a human.
What does it cost?
Free. Apache-2.0 licensed. You pay only for whichever AI agent you use (Claude, ChatGPT, Codex, etc.), and that's billed by your AI provider, not by us.
Safety model
| Tier | Examples | Recommended agent policy |
|---|
| Read | fleet health, fleet down, fleet renewals, fleet changes, fleet find, fleet fanout (except at a secret-bearing path, including configurationChanges and apiRequests), fleet admins, fleet flaps, fleet blame, fleet push without --apply (current credential fields print as [redacted]), every GET except the secret-bearing ones | Allow |
| Credential (incl. secret-returning reads) | auth set-token, administered generate-identities-me-api-keys, admin and SAML writes, organizations login-security update-organization (IP ranges and two-factor: can lock every admin and API key out), organizations update (api.enabled=false cuts every API key off), and reads that return live secrets: networks wireless get-network-ssids (Wi-Fi PSK), networks wireless get-network-ssid-identity-psks, networks snmp get-network, organizations snmp get-organization, organizations appliance get-organization-vpn-third-party-vpnpeers, organizations sm get-organization-vpp-accounts (Apple VPP token), organizations configuration-changes get-organization (the raw change log carries API request bodies; fleet changes is the redacted view), organizations action-batches get-organization-action-batch (returns each pushed action's body), networks webhooks get-network-payload-templates (header templates can hold an Authorization value), networks webhooks get-network-http-servers and organizations webhooks get-organization-logs, alert profile reads such as organizations alerts get-organization-profiles (an incoming-webhook URL, or an HTTP server ID that encodes it, is a credential), organizations api-requests get-organization (request paths can carry those IDs), license-key reads such as organizations licenses get-organization, and the commands that return a time-limited link (organizations devices generate-organization-packet-capture-capture-download-url, devices camera generate-device-snapshot, devices camera clip-device, networks floor-plans get-network, organizations branding-policies get-organization, and the floor-plan, branding-policy and custom-analytics-artifact writes that echo one) | Human-in-the-loop only - an answer carrying a documented secret field (such as psk, secret or communityString) is withheld whole, and the rest of these reads are not redacted; governance.md has the full list |
| Write (routine) | networks update, networks wireless update-network-ssid, networks appliance update-network-vlan, devices update (a write that removes anything, replaces a membership list, changes who can sign in or call the API, or affects power, firmware or connectivity is never routine, and neither is any POST ending in an action such as /bind, /move or /remove - see governance.md) | Preview with --dry-run, then a reviewed write |
| Bulk and device-disrupting | fleet push --apply, organizations action-batches create-organization-action-batch, networks devices vmx-network-claim, devices reboot device, devices switch cycle-device-ports, networks firmware-upgrades update-network, networks firmware-upgrades create-network-rollback, the staged firmware event writes (networks firmware-upgrades create-network-staged-event and friends) devices sensor create-device-command (MT40 power cycle), and networks switch create-network-stack / add-network-stack (forming a stack reboots the switches) | Human-in-the-loop only |
| Destructive | every delete, networks devices remove-network, organizations inventory release-from-organization, networks sm wipe-network-devices, networks switch update-network-stack (a members list replaces the whole stack), organizations wireless create-organization-zigbee-disenrollment, every .../bulkDelete and .../batchDelete POST | Human-in-the-loop only, explicit confirmation |
The strongest control is the Dashboard admin behind the API key - the CLI can only do what that admin is permitted to do, so generate the key from a read-only admin for reporting. For an MCP agent that should never write, also set CISCO_MERAKI_MCP_READ_ONLY=1: the server then refuses every non-GET call and fleet push --apply. Full details, including how to lock it down, are in governance.md.
Status
Beta. Validated against the Cisco Meraki API surface and being validated with MSPs running it live against their own production tenant in our weekly Build Sessions. RSVP at compoundingteams.com/build-sessions.
Standards. Conforms to the open Agent Skills spec (Anthropic, Dec 2025; 40+ agents). MCP-compatible - works with any MCP-capable agent including Hermes. OpenClaw-ready (frontmatter pre-wired, awaiting OpenClaw launch).
Maintained by Servosity. Apache-2.0 licensed. Built with CLI Printing Press. Last updated: 2026-10-13.