Cisco Umbrella MCP

Cisco Umbrella + AI - for ChatGPT, Claude, GitHub Copilot, Microsoft 365 Copilot, Gemini, and any agent that speaks MCP

Unofficial. Community-built Claude Code Skill and MCP server for the Cisco Umbrella API. Not affiliated with, endorsed by, or sponsored by Cisco Systems, Inc.

[Cisco Umbrella demo - animated preview]

_{▶ Watch the 30-second demo - demo data is simulated; every command shown exists in the real CLI.}

Every Cisco Umbrella v2 API call, plus MSP child-org switching, domain verdicts, list diffs, and fleet health. Works with the AI you already use - ChatGPT (Plus/Pro+), Claude Desktop, Codex, Claude Code, Claude Cowork, and GitHub Copilot - plus Microsoft 365 Copilot / Copilot Studio and Google Gemini via the remote path. 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):

For ChatGPT, run cisco-umbrella-mcp --transport http with CISCO_UMBRELLA_MCP_HTTP_TOKEN set, on an HTTPS host; every request sends that token. See mcp-install.md.

Also for the Microsoft and Google stacks

Big install base, but an honest heads-up: these are the remote / enterprise path, not the local binary you just installed.

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 Umbrella MCP (.mcpb) - then open Claude Desktop > Settings > Extensions and select the file. One click, no JSON, no shell. (Browse every Cisco Umbrella 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-umbrella@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 Umbrella 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-umbrella/install.sh). If it uses Windows PowerShell, run iwr -useb https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/cisco-umbrella/install.ps1 | iex. Then authenticate per the README and run cisco-umbrella-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-umbrella/install.ps1 | iex

macOS / Linux:

bash <(curl -fsSL https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/cisco-umbrella/install.sh)

The installer drops both cisco-umbrella-cli (the CLI) and cisco-umbrella-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-umbrella-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-umbrella/install.sh)

Windows (PowerShell):

iwr -useb https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/cisco-umbrella/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-umbrella@msp-skills.

Add to Claude Desktop, GitHub Copilot, Gemini CLI, Microsoft 365 Copilot, 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 remote Microsoft 365 Copilot / Copilot Studio path. 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-umbrella --force

Inside a Hermes chat session:

/skills install servosity/msp-skills/skills/cisco-umbrella --force

Hermes speaks MCP natively, so it can also use the cisco-umbrella-mcp server directly - same install path, same env vars.

Install for OpenClaw

Tell your OpenClaw agent (copy this):

Install the cisco-umbrella skill from https://github.com/servosity/msp-skills/tree/main/skills/cisco-umbrella. The skill defines how its required CLI (cisco-umbrella-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

Create an API key in the Umbrella dashboard under Admin > API Keys (in an MSP or Multi-org console, create it in the parent console), give it the scopes you need, then set the key and secret:

macOS / Linux:

export CISCO_UMBRELLA_API_KEY=<api-key>export CISCO_UMBRELLA_API_SECRET=<api-secret>cisco-umbrella-cli doctor

Windows (PowerShell):

$env:CISCO_UMBRELLA_API_KEY = "<api-key>"$env:CISCO_UMBRELLA_API_SECRET = "<api-secret>"cisco-umbrella-cli doctor

To work inside one customer from a parent console key, add --org <child org ID> to any command or set CISCO_UMBRELLA_ORG_ID; cisco-umbrella-cli managed-customers list shows the IDs.

doctor checks config, paths and API reachability, then proves the credential with one read of /deployments/v2/networks. A key without that scope reads as scope-limited, which still means the key was accepted.

What this skill does

hygiene and search read the local store: run cisco-umbrella-cli sync first (add --org <child id> to sync one customer into its own store).

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 Cisco Umbrella integrations and MCP servers proxy each question into a live API call. That's fine for one record. It dies at scale, when you're asking "which of my 40 customers have roaming computers that stopped checking in?" and the answer means switching into 40 child orgs.

This skill syncs Cisco Umbrella into a local SQLite mirror with full-text search. Aggregate questions become one local SQL join: instant, offline, and the AI sees the answer, not the raw data. Compound commands like fleet health (a child-org token per customer, then roaming computers, virtual appliances and tunnels rolled into one row each), verdict (Investigate, every destination list including parent-domain entries, and the activity report) and allowlist audit (allow lists checked against Investigate and against your block lists) join across customers, lists and reports - work a stateless API wrapper can't do.

The pain this closes

In an Umbrella MSP or Multi-org console, deployment status lives inside each child org. Cisco's own description is that the console lists every child org so it is "easy to switch in and out between customers" (Cisco partner blog, December 2022). Switching in is easy; doing it for every customer to find roaming computers that are off or not checking in, a virtual appliance that is not healthy or a tunnel that is down, is a morning of clicking. Triage has the same shape: a domain's category and risk score live in Investigate, its list membership in every destination list, and who resolved it in an activity report.

• cisco-umbrella-cli fleet health - one row per managed customer, from a parent console key.
• cisco-umbrella-cli verdict <domain> - the whole answer for a domain in one call.
• cisco-umbrella-cli allowlist audit - allow-list entries that have turned risky or conflict with a block list.
• cisco-umbrella-cli destination-lists diff <id> --file <blocklist> - the plan to make a list match your file; --apply makes it so.

See pain-point.md for the longer narrative.

Frequently asked questions

Does this work with ChatGPT?

Yes, on Plus, Pro, Team, Business, Enterprise, and Education plans (Free tier does not yet expose Developer Mode). ChatGPT connects to remote MCP servers over HTTPS, not local stdio binaries. The Cisco Umbrella MCP server runs locally over stdio by default. Its HTTP mode (cisco-umbrella-mcp --transport http) serves http://127.0.0.1:7777/mcp, refuses to start until CISCO_UMBRELLA_MCP_HTTP_TOKEN is set, requires Authorization: Bearer <token> on every request, and needs --tls-cert/--tls-key for any other address. See mcp-install.md. Step-by-step in mcp-install.md.

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 Umbrella credentials once.

Is my Cisco Umbrella 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, and are never bundled into this repo or transmitted anywhere by MSP Skills. The first API call with them saves the key, the secret and the bearer token to credentials.toml in the data directory (doctor prints it): owner-only, not encrypted. Later runs use them after you unset the variables, until auth logout deletes the file; while the variables stay set, the next API call saves them again. A key named on the run itself (CISCO_UMBRELLA_API_KEY, or the config file the run loads) wins over a different key, bearer or auth set-token token saved by an earlier run, so naming another customer's key reaches that customer.

Does it work with an MSP or Multi-org console?

Yes. Create the API key in the parent console. Every command takes --org <child org ID> (or CISCO_UMBRELLA_ORG_ID) and mints a token for that customer with the X-Umbrella-OrgId header; fleet health walks every managed customer for you. The child token stays in memory, and each customer's synced data goes to its own local store.

Which API key scopes does it need?

Only the scopes for what you run: Reports for activity, report-card and the resolver lookup in verdict; Deployments for networks, roaming computers and fleet health, which also needs the Admin customers scope; Policies for destination lists; Investigate for categorization and risk scores; Admin for keys and users. A key missing a scope gets a 403. Read-only scopes cannot change anything.

Will this hit the Umbrella API rate limits?

It is built to stay inside them. Token requests are limited to 20 a minute, so fleet health starts customers 3 seconds apart; narrow a large fleet with --customers or --max-customers. Sync once, and search, hygiene and analytics answer from the local store without new API calls.

Why can't it create or refresh an API key?

Umbrella returns a newly generated secret only once and cannot show it again, and this program never prints a credential it holds or receives, so the secret never lands in an agent transcript or a log. Together that would destroy the only copy, and a refresh would also break whatever uses the key now. So five actions are refused here, before anything is sent, over the CLI and MCP alike: api-keys create, api-keys refresh, s3-bucket-keys, tunnels add with no secret of your own, and tunnels update-credentials with auto-rotate or with no secret of your own. Do them in the Umbrella dashboard, where the secret is shown. --dry-run still previews them. Reads stay: api-keys list and api-keys get print the record with every credential-named field (clientSecret, clientId and the like) replaced by [redacted]. Any other output that repeats a credential the program holds, plain or encoded, is replaced whole by [credkit: withheld from <sink>: it carried a credential this program holds], where <sink> names where it was going.

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

Over MCP, every write to Umbrella returns a preview until the call carries confirm: true, a preview changes nothing on this machine, local learn-store tools (teach, learnings_*, playbook_*) run with no preview and change only this machine, and CISCO_UMBRELLA_MCP_READ_ONLY=1 removes writes entirely.

The strongest control is the scope you grant the Cisco Umbrella credentials - the CLI can only do what the credentials are permitted to do. Full details, including how to lock it down, are in governance.md.

Status

Beta. Validated against the Cisco Umbrella 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: 2026-10-29.