---

name: frontmcp-deployment description: 'Use when deploying, building for production, packaging, or shipping a FrontMCP server. Covers build targets (node, cli SEA binary, browser, embeddable SDK, mcpb archive for Claude Desktop, serverless) and deploying to Vercel (with Vercel KV), AWS Lambda (API Gateway, SAM, CDK), Cloudflare Workers (KV, D1, Durable Objects, v1.3 skills-only), and Node (multi-stage Docker, docker-compose, PM2, nginx). Also the frontmcp.deploy.yaml manifest plus GitHub Action push-resync, MCP client integration / .mcp.json for Claude Desktop, Claude Code, Cursor, and VS Code over stdio or HTTP, MCP protocol revisions (serving 2026-07-28 alongside 2024-11-05 through 2025-11-25), and listing a server in the MCP Registry (domain proof, SEP-2127 server card, AI Catalog). Triggers: deploy, build for production, dockerize, containerize, serverless, edge runtime, go live, ship it.' tags: [router, deployment, node, vercel, lambda, cloudflare, cli, browser, sdk, guide, protocol, mcp-20260728, mcp-registry] category: deployment targets: [all] bundle: [recommended, minimal, full] priority: 10 visibility: both license: Apache-2.0 metadata: docs: https://frontmcp.dev/learn/running-frontmcp-anywhere

FrontMCP Deployment Router

Entry point for deploying and building FrontMCP servers. This skill helps you choose the right deployment target or build format based on your infrastructure requirements.

When to Use This Skill

Must Use

• Choosing between deployment targets (Node vs Vercel vs Lambda vs Cloudflare) for a new project
• Deciding on a build format (server vs CLI vs browser vs SDK) for distribution
• Planning infrastructure and need to understand trade-offs between deployment options

Recommended

• Comparing serverless platforms for cost, cold-start, and feature support
• Understanding which transport protocol and storage provider each target requires
• Migrating from one deployment target to another

Skip When

• You already know your deployment target (go directly to deploy-to-node, deploy-to-vercel, etc.)
• You need to configure server settings, not deploy (see frontmcp-config)
• You need to build components, not ship them (see frontmcp-development)

Decision: Use this skill when you need to figure out WHERE to deploy. Open the matching reference under references/ directly when you already know.

Prerequisites

• A working FrontMCP server with at least one @App and one @Tool (see frontmcp-development)
• Server configuration completed (see frontmcp-config)
• Tests passing locally (see frontmcp-testing)

Steps

1. Review the Scenario Routing Table and Target Comparison below to choose a deployment target
2. Run frontmcp build --target <target> to produce the build output
3. Follow the specific deployment reference (e.g., references/deploy-to-node.md, references/deploy-to-vercel.md) for platform instructions
4. Verify with the Post-Deployment checklist at the end of this skill

Scenario Routing Table

Scenario	Reference	Description
Long-running server on VPS, Docker, or bare metal	deploy-to-node	Node.js with stdio or HTTP transport, PM2/Docker for process management
Serverless with zero config and Vercel KV	deploy-to-vercel	Vercel Functions with Streamable HTTP, Vercel KV for storage
AWS serverless with API Gateway	deploy-to-lambda	Lambda + API Gateway with Streamable HTTP, DynamoDB or ElastiCache
Edge computing with global distribution	deploy-to-cloudflare	Cloudflare Workers with KV or Durable Objects for storage
Hosted FrontMCP (v1.3 skills-only model)	deploy-to-cloudflare-skills-only	Cloudflare Worker as the MCP control plane; OpenAPI is capability inventory; agent uses 4 meta-tools + AgentScript; hot-reload via GitHub Action
Author frontmcp.deploy.yaml	deploy-manifest-yaml	v1 schema reference: runtime, server, specs, skills, tags, classification, bindings, signing, auth, secrets, environments
Standalone executable binary for distribution	build-for-cli	Single-binary CLI with stdio transport, embedded storage
Run MCP in a web browser	build-for-browser	Browser-compatible bundle with in-memory transport
Embed MCP into an existing Node.js application	build-for-sdk	Library build for programmatic usage without standalone server
Write a Dockerfile for Node.js deployment	deploy-to-node-dockerfile	Dockerfile configuration for Node.js deployment
Configure Vercel-specific settings (vercel.json)	deploy-to-vercel-config	Vercel-specific configuration (vercel.json)
Connect MCP clients (Claude, Cursor, VS Code)	mcp-client-integration	Configure .mcp.json for stdio, HTTP, or Unix socket transport
Serve or consume MCP protocol 2026-07-28	protocol-versions	Stateless requests, server/discover, mirrored headers, MRTR, tasks extension, and the McpStatelessClient
List the server in the MCP Registry	list-in-mcp-registry	@FrontMcp({ registry }): domain proof, SEP-2127 server card and AI Catalog; frontmcp registry keygen / publish / status

CLI Commands for Deployment and Operations

Beyond frontmcp build, the CLI provides commands for the full deployment lifecycle:

Command	Description
frontmcp build -t <target>	Build for target: node, vercel, lambda, cloudflare, cli, browser, sdk
frontmcp build -t cli --js	Build CLI as JS bundle (instead of native binary via SEA)
frontmcp build --no-clean	Keep existing output instead of clearing the target's output directory first
frontmcp start <name>	Start a named MCP server with supervisor (process management)
frontmcp stop <name>	Stop managed server (-f for force kill)
frontmcp restart <name>	Restart managed server
frontmcp status [name]	Show process status (detail if name given, table if omitted)
frontmcp list	List all managed processes
frontmcp logs <name>	Tail log output (-F follow, -n lines)
frontmcp socket <entry>	Start Unix socket daemon for local MCP server
frontmcp service <action>	Install/uninstall systemd (Linux) or launchd (macOS) service
frontmcp install <source>	Install MCP app from npm, local path (., ./app, /abs/app), git, or a project's dist/ / dist/node (dist/node wins over other targets; installs the external runtime packages (vectoriadb comes with the SDK) and the optional @frontmcp/sdk and @frontmcp/utils peers the project declares, those in optionalDependencies as optional; frontmcp start <name> then runs it from its install dir)
frontmcp uninstall <name>	Remove installed MCP app
frontmcp configure <name>	Re-run setup questionnaire for installed app
frontmcp doctor	Check Node.js/npm versions and tsconfig requirements
frontmcp inspector	Launch MCP Inspector for debugging
frontmcp init	Create or fix tsconfig.json for FrontMCP (edits in place, keeps comments; never overwrites a file it cannot parse or that declares a required option twice)
frontmcp registry keygen	Create the Ed25519 key for the MCP Registry's domain proof; prints MCP_REGISTRY_PUBLIC_KEY and the DNS TXT value
frontmcp registry publish	Build and validate server.json, then log in and publish it to the MCP Registry (--dry-run previews)
frontmcp registry status	Mark one version (--server-version) or every version (--all-versions) active, deprecated or deleted

Target Comparison

Target	Transport	Storage	Cold Start	Stateful	Best For
Node	stdio, SSE, Streamable HTTP	Redis, SQLite, memory	None	Yes	Full-featured production servers
Vercel	Streamable HTTP (stateless)	Vercel KV	~250ms	No	Rapid deployment, hobby/startup
Lambda	Streamable HTTP (stateless)	DynamoDB, ElastiCache	~500ms	No	AWS ecosystem, event-driven
Cloudflare	Streamable HTTP (stateless)	KV, Durable Objects	~5ms	Limited	Edge-first, global latency
CLI	stdio	SQLite, memory	None	Yes	Desktop tools, local agents
Browser	In-process direct client	memory	None	Yes	Client-side AI, demos
SDK	Programmatic	Configurable	None	Yes	Embedding in existing apps

Note on storage: Only redis and vercel-kv are SDK-native providers. DynamoDB, Cloudflare KV, D1, and Durable Objects are platform-side — wire them in your tools using the platform SDK / Workers bindings. The Cloudflare build adapter actively rejects redis: { ... } and sqlite: { ... } configs at build time because Workers has no Node TCP / fs.

Note on browser: "In-process direct client" means an in-memory DirectClient created via connect()/create() from @frontmcp/sdk. There is no separate "in-memory transport" — the client and server share the same JS heap.

Cross-Cutting Patterns

Pattern	Rule
Transport selection	Stateful servers (Node, CLI) can use stdio or SSE; serverless must use Streamable HTTP (stateless)
Storage mapping	SDK-native: redis (Node, Lambda+ElastiCache, CLI), vercel-kv (Vercel, can also work on Cloudflare via Upstash HTTP). Platform-side (you wire it in tools): DynamoDB, Cloudflare KV/D1/DO. Cloudflare build rejects redis/sqlite configs.
Environment variables	Never hardcode secrets; use .env locally, platform secrets in production
Build command	All targets: frontmcp build --target <target> produces optimized output
Entry point	All targets require export default of the @FrontMcp class from main.ts

Common Patterns

Pattern	Correct	Incorrect	Why
Target selection	Choose based on infrastructure constraints	Choose based on familiarity	Each target has different transport, storage, and cold-start characteristics
Serverless storage	Use platform-native storage (Vercel KV, DynamoDB)	Use Redis on serverless	Platform-native storage avoids VPC/connection overhead on cold starts
Environment config	Platform secrets (Vercel env, AWS SSM)	.env files in production	Platform secrets are encrypted, rotatable, and not committed to git
Build verification	Run frontmcp build --target <target> before deploying	Deploy source code directly	Build step validates config, bundles dependencies, and optimizes output

Verification Checklist

Pre-Deployment

• frontmcp build --target <target> completes without errors
• Environment variables configured for the target platform
• Storage provider configured and accessible (Redis, KV, DynamoDB, etc.)
• Transport protocol matches target requirements (stateless for serverless)

Post-Deployment

• Health check endpoint responds
• tools/list returns expected tools
• Tool execution works end-to-end
• Storage persistence verified (create, read, restart, read again)

Troubleshooting

Problem	Cause	Solution
Cold start timeout on serverless	Bundle too large or heavy initialization	Lazy-load providers; reduce bundle with tree shaking; increase function timeout
Session lost between requests	Using memory storage on stateless serverless	Switch to platform-native storage (Vercel KV, DynamoDB, etc.)
CORS errors on browser/web clients	HTTP CORS not configured	Add CORS config via configure-http skill
Build fails with missing module	Node-only module in browser/edge build	Use conditional imports or @frontmcp/utils cross-platform utilities
TS2688: Cannot find type definition file for 'node' under Yarn	An older CLI shelled out to npx tsc, which starts a process that never loads .pnp.cjs	Upgrade the CLI — the build now runs the project's own typescript with the current Node binary, or delegates to yarn/pnpm/bun when there is no local install
@frontmcp/sdk tried to access <pkg> (a peer dependency)	Yarn Plug'n'Play enforces peer dependencies that a hoisted node_modules tree tolerates	Add nodeLinker: node-modules to .yarnrc.yml and reinstall (frontmcp create now scaffolds this for Yarn projects)
Deleted source file still present in dist/	An older CLI never cleared the output directory, or --no-clean was passed	Rebuild without --no-clean; the build now clears the target's output directory first

Examples

Each reference has matching examples under examples/<reference>/:

build-for-browser

Example	Level	Description
browser-build-with-custom-entry	Intermediate	Build a browser bundle using a dedicated client entry file that avoids Node.js-only imports.
browser-crypto-and-storage	Advanced	Use @frontmcp/utils crypto functions (WebCrypto API) and in-memory storage in browser environments.
react-provider-setup	Basic	Connect a React application to a remote FrontMCP server using @frontmcp/react.

build-for-cli

Example	Level	Description
cli-binary-build	Basic	Build a FrontMCP server as a standalone binary using Node.js Single Executable Applications (SEA).
unix-socket-daemon	Intermediate	Run a FrontMCP server as a local daemon accessible via Unix socket for IDE extensions and local MCP clients.

build-for-mcpb

Example	Level	Description
mcpb-bundle-build	Basic	Produce a .mcpb archive for Claude Desktop with metadata, tools, and install-time user_config.

build-for-sdk

Example	Level	Description
connect-openai	Intermediate	Use connectOpenAI() to get tools formatted for OpenAI's function-calling API.
create-flat-config	Basic	Spin up an in-memory FrontMCP server from a flat config object using create().
multi-platform-connect	Advanced	Connect the same FrontMCP server to multiple LLM platforms using platform-specific connect*() functions.

deploy-to-cloudflare

Example	Level	Description
basic-worker-deploy	Basic	Deploy a FrontMCP server to Cloudflare Workers with a minimal configuration.
worker-custom-domain	Advanced	Scaffold a FrontMCP project targeting Cloudflare, configure a custom domain, and verify the deployment.
worker-with-kv-storage	Intermediate	Deploy a FrontMCP server to Cloudflare Workers with KV namespace for session and state storage.

deploy-to-lambda

Example	Level	Description
cdk-deployment	Advanced	Deploy a FrontMCP server to AWS Lambda using CDK with provisioned concurrency and secrets management.
lambda-handler-with-cors	Intermediate	Create a custom Lambda handler with an explicit API Gateway definition for CORS support.
sam-template-basic	Basic	Deploy a FrontMCP server to AWS Lambda with API Gateway using a SAM template.

deploy-to-node-dockerfile

Example	Level	Description
basic-multistage-dockerfile	Basic	A minimal multi-stage Dockerfile for building and running a FrontMCP server in production.
secure-nonroot-dockerfile	Advanced	A production Dockerfile with a non-root user, proper ownership, and security hardening.

deploy-to-node

Example	Level	Description
docker-compose-with-redis	Basic	Deploy a FrontMCP server with Redis using Docker Compose for production.
pm2-with-nginx	Intermediate	Deploy a FrontMCP server on bare metal using PM2 for process management and NGINX for TLS termination.
resource-limits	Advanced	Configure resource limits, health checks, and environment variables for a production FrontMCP deployment.

deploy-to-vercel-config

Example	Level	Description
minimal-vercel-config	Basic	The minimum vercel.json needed to deploy a FrontMCP server to Vercel.
vercel-config-with-security-headers	Intermediate	A complete vercel.json with per-route security headers for health, MCP, and all other endpoints.

deploy-to-vercel

Example	Level	Description
vercel-mcp-endpoint-test	Advanced	Verify a Vercel-deployed FrontMCP server by testing health, tool listing, and tool invocation.
vercel-with-kv	Basic	Deploy a FrontMCP server to Vercel serverless functions with Vercel KV for session persistence.
vercel-with-skills-cache	Intermediate	Deploy a FrontMCP server to Vercel with skills enabled and KV-backed skill caching.

mcp-client-integration

Example	Level	Description
stdio-npx	Basic	Publish a FrontMCP server to npm and configure MCP clients to use it with npx --stdio.
http-remote	Basic	Connect an MCP client to a FrontMCP server running as an HTTP server, locally or remotely.
stdio-binary-with-env	Intermediate	Configure a local FrontMCP CLI binary with environment variables and custom arguments in MCP client configs.

Accessing This Skill

Skills are distributed as plain SKILL.md files plus a sibling references/ and examples/ tree, so consumers can pick whichever access mode fits:

Mode	How it works
Filesystem	Read libs/skills/catalog/frontmcp-deployment/ directly from a clone of the catalog repo, or from a published @frontmcp/skills install. SKILL.md is the entry point.
frontmcp CLI	frontmcp skills list, frontmcp skills read frontmcp-deployment, frontmcp skills read frontmcp-deployment:references/<file>.md, frontmcp skills install frontmcp-deployment — no server required.
MCP skill://	When a developer mounts this skill into their own FrontMCP server (@FrontMcp({ skills: [...] })), the SDK exposes it via SEP-2640 resources: skill://frontmcp-deployment/SKILL.md, skill://frontmcp-deployment/references/{file}.md, etc. The server’s skill://index.json returns the SEP-2640 discovery document for everything mounted on it.

The catalog itself is not an MCP server. The skill:// URIs only resolve when a server has been configured to host this skill.

Reference

• Runtime Modes
• Related skills: deploy-to-node, deploy-to-vercel, deploy-to-lambda, deploy-to-cloudflare, build-for-cli, build-for-browser, build-for-sdk, configure-transport