$ loading_
Helps developers plan the right MCP server architecture and build approach.
Copy the install command and let the AI configure it · recommended for beginners
Please install the "build-mcp-server" skill from askskill: 1. Download https://raw.githubusercontent.com/anthropics/claude-plugins-official/main/plugins/mcp-server-dev/skills/build-mcp-server/SKILL.md 2. Save it as ~/.claude/skills/build-mcp-server/SKILL.md 3. Reload skills and tell me it's ready
I want to wrap our company REST API as an MCP server for Claude. First, determine whether I should use remote HTTP, MCPB, or local stdio based on users, auth, number of actions, and UI needs, then suggest a tool design pattern.
It first clarifies requirements, then recommends a deployment model, auth approach, and tool design pattern.
I want Claude to connect to a local filesystem and desktop app for internal team use only. Help me decide whether MCPB or local stdio fits better, and explain the reasons and next development path.
It provides a deployment recommendation for local scenarios, including fit, constraints, and a development direction.
I need to connect a platform with hundreds of endpoints to Claude. Based on the number of actions, help me choose between one-tool-per-action and a search-plus-execute pattern, and explain why.
It recommends a suitable tool pattern for wrapping a large API surface and explains the tradeoffs.
When a developer is about to build an MCP server, this skill helps clarify what it connects to, who will use it, auth needs, and interaction requirements before coding. It helps choose between remote HTTP, MCPB, or local stdio.
When a team wants to expose a SaaS, REST, or GraphQL service to Claude, this skill acts as the entry point. It recommends a deployment model first, then suggests how the tools should be designed to reduce rework later.
If a tool needs mid-call user input or richer visual display, this skill helps distinguish between elicitation, plain text/JSON responses, or moving to MCP app widgets.
The README excerpt focuses on planning an MCP server before writing code. It emphasizes discovery first: understanding what the server connects to, who will use it, how many actions it exposes, whether it needs mid-call input or rich UI, and what auth the upstream service uses. Based on those answers, it guides the developer toward an opinionated deployment choice such as remote HTTP, MCPB, or local stdio, along with an appropriate tool design pattern.
You are guiding a developer through designing and building an MCP server that works seamlessly with Claude. MCP servers come in many forms — picking the wrong shape early causes painful rewrites later. Your first job is discovery, not code.
Load Claude-specific context first. The MCP spec is generic; Claude has additional auth types, review criteria, and limits. Before answering questions or scaffolding, fetch https://claude.com/docs/llms-full.txt (the full export of the Claude connector docs) so your guidance reflects Claude's actual constraints.
Do not start scaffolding until you have answers to the questions in Phase 1. If the user's opening message already answers them, acknowledge that and skip straight to the recommendation.
Ask these questions conversationally (batch them into one message, don't interrogate one-at-a-time). Adapt wording to what the user has already told you.
| If it connects to… | Likely direction |
|---|---|
| A cloud API (SaaS, REST, GraphQL) | Remote HTTP server |
| A local process, filesystem, or desktop app | MCPB or local stdio |
| Hardware, OS-level APIs, or user-specific state | MCPB |
| Nothing external — pure logic / computation | Either — default to remote |
This determines the tool-design pattern — see Phase 3.
references/elicitation.md.@modelcontextprotocol/ext-apps. See build-mcp-app skill.references/auth.mdBased on the answers, recommend one path. Be opinionated. The ranked options:
A hosted service speaking MCP over streamable HTTP. This is the recommended path for anything wrapping a cloud API.
Why it wins:
Choose this unless the server must touch the user's local machine.
→ Fastest deploy: Cloudflare Workers — references/deploy-cloudflare-workers.md (zero to live URL in two commands)
→ Portable Node/Python: references/remote-http-scaffold.md (Express or FastMCP, runs on any host)
If a tool just needs the user to confirm, pick an option, or fill a short form, elicitation does it with zero UI code. The server sends a flat JSON schema; the host renders a native form. Spec-native, no extra packages.
…
It is an entry-point skill for building MCP servers, focused on discovery rather than jumping straight into code. It recommends a deployment model and tool design direction based on what you connect to, who will use it, action count, interaction needs, and auth.
The excerpt indicates that wrapping a cloud API usually points to remote HTTP, while local processes, filesystems, desktop apps, hardware, or user-specific local state lean toward MCPB or local stdio. For internal team-only use, local stdio can be an easier prototyping option.
No. The excerpt explicitly says the first job is discovery, and scaffolding should not start until the key Phase 1 questions are answered.
Package a local MCP server into a runtime-bundled .mcpb file.
Browse Anthropic’s official directory of quality Claude Code plugins.
Create Claude Code plugin hooks for tool validation and event-driven automation.
Build single-file interactive HTML playgrounds with live preview and prompt export.
Create and structure reusable slash commands for Claude Code workflows.
This skill should be used when the user asks about "plugin settings", "store plugin configuration", "user-configurable plugin", ".local.md files", "plugin state files", "read YAML frontmatter", "per-project plugin settings", or wants to make plugin behavior configurable. Documents the .claude/plugin-name.local.md pattern for storing plugin-specific configuration with YAML frontmatter and markdown content.
Build high-quality MCP servers that connect LLMs with external APIs safely.
Bootstrap MCP servers quickly and extend tools with a built-in CLI.
Build MCP servers in TypeScript with less boilerplate and testing helpers.
Build local MCP servers and AI tools from plain-language descriptions.
Connect MCP clients to basic tools and customizable prompts.
Return personalized greetings to demonstrate basic MCP tool integration.