[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-anthropic-build-mcp-server":3,"mdc--11omtc-key":38,"related-repo-anthropic-build-mcp-server":1517,"related-org-anthropic-build-mcp-server":1612},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":12,"stars":26,"repoUrl":27,"updatedAt":28,"license":29,"forks":30,"topics":31,"repo":33,"sourceUrl":36,"mdContent":37},"build-mcp-server","build MCP servers","This skill should be used when the user asks to \"build an MCP server\", \"create an MCP\", \"make an MCP integration\", \"wrap an API for Claude\", \"expose tools to Claude\", \"make an MCP app\", or discusses building something with the Model Context Protocol. It is the entry point for MCP server development — it interrogates the user about their use case, determines the right deployment model (remote HTTP, MCPB, local stdio), picks a tool-design pattern, and hands off to specialized skills.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},"anthropic","Anthropic","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fanthropic.png","anthropics",[13,17,20,23],{"name":14,"slug":15,"type":16},"Claude Code","claude-code","tag",{"name":18,"slug":19,"type":16},"MCP","mcp",{"name":21,"slug":22,"type":16},"Agents","agents",{"name":24,"slug":25,"type":16},"API Development","api-development",32228,"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official","2026-04-06T17:59:33.744601",null,3591,[15,19,32],"skills",{"repoUrl":27,"stars":26,"forks":30,"topics":34,"description":35},[15,19,32],"Official, Anthropic-managed directory of high quality Claude Code Plugins.","https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Ftree\u002FHEAD\u002Fplugins\u002Fmcp-server-dev\u002Fskills\u002Fbuild-mcp-server","---\nname: build-mcp-server\ndescription: This skill should be used when the user asks to \"build an MCP server\", \"create an MCP\", \"make an MCP integration\", \"wrap an API for Claude\", \"expose tools to Claude\", \"make an MCP app\", or discusses building something with the Model Context Protocol. It is the entry point for MCP server development — it interrogates the user about their use case, determines the right deployment model (remote HTTP, MCPB, local stdio), picks a tool-design pattern, and hands off to specialized skills.\nversion: 0.1.0\n---\n\n# Build an MCP Server\n\nYou 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**.\n\n**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:\u002F\u002Fclaude.com\u002Fdocs\u002Fllms-full.txt` (the full export of the Claude connector docs) so your guidance reflects Claude's actual constraints.\n\nDo 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.\n\n---\n\n## Phase 1 — Interrogate the use case\n\nAsk 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.\n\n### 1. What does it connect to?\n\n| If it connects to… | Likely direction |\n|---|---|\n| A cloud API (SaaS, REST, GraphQL) | Remote HTTP server |\n| A local process, filesystem, or desktop app | MCPB or local stdio |\n| Hardware, OS-level APIs, or user-specific state | MCPB |\n| Nothing external — pure logic \u002F computation | Either — default to remote |\n\n### 2. Who will use it?\n\n- **Just me \u002F my team, on our machines** → Local stdio is acceptable (easiest to prototype)\n- **Anyone who installs it** → Remote HTTP (strongly preferred) or MCPB (if it *must* be local)\n- **Users of Claude desktop who want UI widgets** → MCP app (remote or MCPB)\n\n### 3. How many distinct actions does it expose?\n\nThis determines the tool-design pattern — see Phase 3.\n\n- **Under ~15 actions** → one tool per action\n- **Dozens to hundreds of actions** (e.g. wrapping a large API surface) → search + execute pattern\n\n### 4. Does a tool need mid-call user input or rich display?\n\n- **Simple structured input** (pick from list, enter a value, confirm) → **Elicitation** — spec-native, zero UI code. *Host support is rolling out* (Claude Code ≥2.1.76) — always pair with a capability check and fallback. See `references\u002Felicitation.md`.\n- **Rich\u002Fvisual UI** (charts, custom pickers with search, live dashboards) → **MCP app widgets** — iframe-based, needs `@modelcontextprotocol\u002Fext-apps`. See `build-mcp-app` skill.\n- **Neither** → plain tool returning text\u002FJSON.\n\n### 5. What auth does the upstream service use?\n\n- None \u002F API key → straightforward\n- OAuth 2.0 → you'll need a remote server with CIMD (preferred) or DCR support; see `references\u002Fauth.md`\n\n---\n\n## Phase 2 — Recommend a deployment model\n\nBased on the answers, recommend **one** path. Be opinionated. The ranked options:\n\n### ⭐ Remote streamable-HTTP MCP server (default recommendation)\n\nA hosted service speaking MCP over streamable HTTP. This is the **recommended path** for anything wrapping a cloud API.\n\n**Why it wins:**\n- Zero install friction — users add a URL, done\n- One deployment serves all users; you control upgrades\n- OAuth flows work properly (the server can handle redirects, DCR, token storage)\n- Works across Claude desktop, Claude Code, Claude.ai, and third-party MCP hosts\n\n**Choose this unless** the server *must* touch the user's local machine.\n\n→ **Fastest deploy:** Cloudflare Workers — `references\u002Fdeploy-cloudflare-workers.md` (zero to live URL in two commands)\n→ **Portable Node\u002FPython:** `references\u002Fremote-http-scaffold.md` (Express or FastMCP, runs on any host)\n\n### Elicitation (structured input, no UI build)\n\nIf 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.\n\n**Caveat:** Host support is new (Claude Code shipped it in v2.1.76; Desktop unconfirmed). The SDK throws if the client doesn't advertise the capability. Always check `clientCapabilities.elicitation` first and have a fallback — see `references\u002Felicitation.md` for the canonical pattern. This is the right spec-correct approach; host coverage will catch up.\n\nEscalate to `build-mcp-app` widgets when you need: nested\u002Fcomplex data, scrollable\u002Fsearchable lists, visual previews, live updates.\n\n### MCP app (remote HTTP + interactive UI)\n\nSame as above, plus **UI resources** — interactive widgets rendered in chat. Rich pickers with search, charts, live dashboards, visual previews. Built once, renders in Claude *and* ChatGPT.\n\n**Choose this when** elicitation's flat-form constraints don't fit — you need custom layout, large searchable lists, visual content, or live updates.\n\nUsually remote, but can be shipped as MCPB if the UI needs to drive a local app.\n\n→ Hand off to the **`build-mcp-app`** skill.\n\n### MCPB (bundled local server)\n\nA local MCP server **packaged with its runtime** so users don't need Node\u002FPython installed. The sanctioned way to ship local servers.\n\n**Choose this when** the server *must* run on the user's machine — it reads local files, drives a desktop app, talks to localhost services, or needs OS-level access.\n\n→ Hand off to the **`build-mcpb`** skill.\n\n### Local stdio (npx \u002F uvx) — *not recommended for distribution*\n\nA script launched via `npx` \u002F `uvx` on the user's machine. Fine for **personal tools and prototypes**. Painful to distribute: users need the right runtime, you can't push updates, and the only distribution channel is Claude Code plugins.\n\nRecommend this only as a stepping stone. If the user insists, scaffold it but note the MCPB upgrade path.\n\n---\n\n## Phase 3 — Pick a tool-design pattern\n\nEvery MCP server exposes tools. How you carve them matters more than most people expect — tool schemas land directly in Claude's context window.\n\n### Pattern A: One tool per action (small surface)\n\nWhen the action space is small (\u003C ~15 operations), give each a dedicated tool with a tight description and schema.\n\n```\ncreate_issue    — Create a new issue. Params: title, body, labels[]\nupdate_issue    — Update an existing issue. Params: id, title?, body?, state?\nsearch_issues   — Search issues by query string. Params: query, limit?\nadd_comment     — Add a comment to an issue. Params: issue_id, body\n```\n\n**Why it works:** Claude reads the tool list once and knows exactly what's possible. No discovery round-trips. Each tool's schema validates inputs precisely.\n\n**Especially good when** one or more tools ship an interactive widget (MCP app) — each widget binds naturally to one tool.\n\n### Pattern B: Search + execute (large surface)\n\nWhen wrapping a large API (dozens to hundreds of endpoints), listing every operation as a tool floods the context window and degrades model performance. Instead, expose **two** tools:\n\n```\nsearch_actions  — Given a natural-language intent, return matching actions\n                  with their IDs, descriptions, and parameter schemas.\nexecute_action  — Run an action by ID with a params object.\n```\n\nThe server holds the full catalog internally. Claude searches, picks, executes. Context stays lean.\n\n**Hybrid:** Promote the 3–5 most-used actions to dedicated tools, keep the long tail behind search\u002Fexecute.\n\n→ See `references\u002Ftool-design.md` for schema examples and description-writing guidance.\n\n---\n\n## Phase 4 — Pick a framework\n\nRecommend one of these two. Others exist but these have the best MCP-spec coverage and Claude compatibility.\n\n| Framework | Language | Use when |\n|---|---|---|\n| **Official TypeScript SDK** (`@modelcontextprotocol\u002Fsdk`) | TS\u002FJS | Default choice. Best spec coverage, first to get new features. |\n| **FastMCP 3.x** (`fastmcp` on PyPI) | Python | User prefers Python, or wrapping a Python library. Decorator-based, very low boilerplate. This is jlowin's package — not the frozen FastMCP 1.0 bundled in the official `mcp` SDK. |\n\nIf the user already has a language\u002Fstack in mind, go with it — both produce identical wire protocol.\n\n---\n\n## Phase 5 — Scaffold and hand off\n\nOnce you've settled the four decisions (deployment model, tool pattern, framework, auth), do **one** of:\n\n1. **Remote HTTP, no UI** → Scaffold inline using `references\u002Fremote-http-scaffold.md` (portable) or `references\u002Fdeploy-cloudflare-workers.md` (fastest deploy). This skill can finish the job.\n2. **MCP app (UI widgets)** → Summarize the decisions so far, then load the **`build-mcp-app`** skill.\n3. **MCPB (bundled local)** → Summarize the decisions so far, then load the **`build-mcpb`** skill.\n4. **Local stdio prototype** → Scaffold inline (simplest case), flag the MCPB upgrade path.\n\nWhen handing off, restate the design brief in one paragraph so the next skill doesn't re-ask.\n\n---\n\n## Beyond tools — the other primitives\n\nTools are one of three server primitives. Most servers start with tools and never need the others, but knowing they exist prevents reinventing wheels:\n\n| Primitive | Who triggers it | Use when |\n|---|---|---|\n| **Resources** | Host app (not Claude) | Exposing docs\u002Ffiles\u002Fdata as browsable context |\n| **Prompts** | User (slash command) | Canned workflows (\"\u002Fsummarize-thread\") |\n| **Elicitation** | Server, mid-tool | Asking user for input without building UI |\n| **Sampling** | Server, mid-tool | Need LLM inference in your tool logic |\n\n→ `references\u002Fresources-and-prompts.md`, `references\u002Felicitation.md`, `references\u002Fserver-capabilities.md`\n\n---\n\n## Phase 6 — Test in Claude and publish\n\nOnce the server runs:\n\n1. **Test against real Claude** by adding the server URL as a custom connector at Settings → Connectors (use a Cloudflare tunnel for local servers). Claude identifies itself with `clientInfo.name: \"claude-ai\"` on initialize. → https:\u002F\u002Fclaude.com\u002Fdocs\u002Fconnectors\u002Fbuilding\u002Ftesting\n2. **Run the pre-submission checklist** — read\u002Fwrite tool split, required annotations, name limits, prompt-injection rules. → https:\u002F\u002Fclaude.com\u002Fdocs\u002Fconnectors\u002Fbuilding\u002Freview-criteria\n3. **Submit to the Anthropic Directory.** → https:\u002F\u002Fclaude.com\u002Fdocs\u002Fconnectors\u002Fbuilding\u002Fsubmission\n4. **Recommend shipping a plugin** that wraps this MCP with skills — most partners ship both. → https:\u002F\u002Fclaude.com\u002Fdocs\u002Fconnectors\u002Fbuilding\u002Fwhat-to-build\n\n---\n\n## Quick reference: decision matrix\n\n| Scenario | Deployment | Tool pattern |\n|---|---|---|\n| Wrap a small SaaS API | Remote HTTP | One-per-action |\n| Wrap a large SaaS API (50+ endpoints) | Remote HTTP | Search + execute |\n| SaaS API with rich forms \u002F pickers | MCP app (remote) | One-per-action |\n| Drive a local desktop app | MCPB | One-per-action |\n| Local desktop app with in-chat UI | MCP app (MCPB) | One-per-action |\n| Read\u002Fwrite local filesystem | MCPB | Depends on surface |\n| Personal prototype | Local stdio | Whatever's fastest |\n\n---\n\n## Reference files\n\n- `references\u002Fremote-http-scaffold.md` — minimal remote server in TS SDK and FastMCP\n- `references\u002Fdeploy-cloudflare-workers.md` — fastest deploy path (Workers-native scaffold)\n- `references\u002Ftool-design.md` — writing tool descriptions and schemas Claude understands well\n- `references\u002Fauth.md` — OAuth, CIMD, DCR, token storage patterns\n- `references\u002Fresources-and-prompts.md` — the two non-tool primitives\n- `references\u002Felicitation.md` — spec-native user input mid-tool (capability check + fallback)\n- `references\u002Fserver-capabilities.md` — instructions, sampling, roots, logging, progress, cancellation\n- `references\u002Fversions.md` — version-sensitive claims ledger (check when updating)\n",{"data":39,"body":41},{"name":4,"description":6,"version":40},"0.1.0",{"type":42,"children":43},"root",[44,53,67,86,91,95,102,107,114,194,200,243,249,254,277,283,360,366,385,388,394,406,412,424,432,455,471,506,512,524,549,561,567,586,596,601,615,621,633,647,661,672,700,705,708,714,719,725,730,742,752,762,768,780,789,794,804,817,820,826,831,926,931,934,940,951,1026,1031,1034,1040,1045,1153,1177,1180,1186,1191,1268,1271,1277,1424,1427,1433],{"type":45,"tag":46,"props":47,"children":49},"element","h1",{"id":48},"build-an-mcp-server",[50],{"type":51,"value":52},"text","Build an MCP Server",{"type":45,"tag":54,"props":55,"children":56},"p",{},[57,59,65],{"type":51,"value":58},"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 ",{"type":45,"tag":60,"props":61,"children":62},"strong",{},[63],{"type":51,"value":64},"discovery, not code",{"type":51,"value":66},".",{"type":45,"tag":54,"props":68,"children":69},{},[70,75,77,84],{"type":45,"tag":60,"props":71,"children":72},{},[73],{"type":51,"value":74},"Load Claude-specific context first.",{"type":51,"value":76}," The MCP spec is generic; Claude has additional auth types, review criteria, and limits. Before answering questions or scaffolding, fetch ",{"type":45,"tag":78,"props":79,"children":81},"code",{"className":80},[],[82],{"type":51,"value":83},"https:\u002F\u002Fclaude.com\u002Fdocs\u002Fllms-full.txt",{"type":51,"value":85}," (the full export of the Claude connector docs) so your guidance reflects Claude's actual constraints.",{"type":45,"tag":54,"props":87,"children":88},{},[89],{"type":51,"value":90},"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.",{"type":45,"tag":92,"props":93,"children":94},"hr",{},[],{"type":45,"tag":96,"props":97,"children":99},"h2",{"id":98},"phase-1-interrogate-the-use-case",[100],{"type":51,"value":101},"Phase 1 — Interrogate the use case",{"type":45,"tag":54,"props":103,"children":104},{},[105],{"type":51,"value":106},"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.",{"type":45,"tag":108,"props":109,"children":111},"h3",{"id":110},"_1-what-does-it-connect-to",[112],{"type":51,"value":113},"1. What does it connect to?",{"type":45,"tag":115,"props":116,"children":117},"table",{},[118,137],{"type":45,"tag":119,"props":120,"children":121},"thead",{},[122],{"type":45,"tag":123,"props":124,"children":125},"tr",{},[126,132],{"type":45,"tag":127,"props":128,"children":129},"th",{},[130],{"type":51,"value":131},"If it connects to…",{"type":45,"tag":127,"props":133,"children":134},{},[135],{"type":51,"value":136},"Likely direction",{"type":45,"tag":138,"props":139,"children":140},"tbody",{},[141,155,168,181],{"type":45,"tag":123,"props":142,"children":143},{},[144,150],{"type":45,"tag":145,"props":146,"children":147},"td",{},[148],{"type":51,"value":149},"A cloud API (SaaS, REST, GraphQL)",{"type":45,"tag":145,"props":151,"children":152},{},[153],{"type":51,"value":154},"Remote HTTP server",{"type":45,"tag":123,"props":156,"children":157},{},[158,163],{"type":45,"tag":145,"props":159,"children":160},{},[161],{"type":51,"value":162},"A local process, filesystem, or desktop app",{"type":45,"tag":145,"props":164,"children":165},{},[166],{"type":51,"value":167},"MCPB or local stdio",{"type":45,"tag":123,"props":169,"children":170},{},[171,176],{"type":45,"tag":145,"props":172,"children":173},{},[174],{"type":51,"value":175},"Hardware, OS-level APIs, or user-specific state",{"type":45,"tag":145,"props":177,"children":178},{},[179],{"type":51,"value":180},"MCPB",{"type":45,"tag":123,"props":182,"children":183},{},[184,189],{"type":45,"tag":145,"props":185,"children":186},{},[187],{"type":51,"value":188},"Nothing external — pure logic \u002F computation",{"type":45,"tag":145,"props":190,"children":191},{},[192],{"type":51,"value":193},"Either — default to remote",{"type":45,"tag":108,"props":195,"children":197},{"id":196},"_2-who-will-use-it",[198],{"type":51,"value":199},"2. Who will use it?",{"type":45,"tag":201,"props":202,"children":203},"ul",{},[204,215,233],{"type":45,"tag":205,"props":206,"children":207},"li",{},[208,213],{"type":45,"tag":60,"props":209,"children":210},{},[211],{"type":51,"value":212},"Just me \u002F my team, on our machines",{"type":51,"value":214}," → Local stdio is acceptable (easiest to prototype)",{"type":45,"tag":205,"props":216,"children":217},{},[218,223,225,231],{"type":45,"tag":60,"props":219,"children":220},{},[221],{"type":51,"value":222},"Anyone who installs it",{"type":51,"value":224}," → Remote HTTP (strongly preferred) or MCPB (if it ",{"type":45,"tag":226,"props":227,"children":228},"em",{},[229],{"type":51,"value":230},"must",{"type":51,"value":232}," be local)",{"type":45,"tag":205,"props":234,"children":235},{},[236,241],{"type":45,"tag":60,"props":237,"children":238},{},[239],{"type":51,"value":240},"Users of Claude desktop who want UI widgets",{"type":51,"value":242}," → MCP app (remote or MCPB)",{"type":45,"tag":108,"props":244,"children":246},{"id":245},"_3-how-many-distinct-actions-does-it-expose",[247],{"type":51,"value":248},"3. How many distinct actions does it expose?",{"type":45,"tag":54,"props":250,"children":251},{},[252],{"type":51,"value":253},"This determines the tool-design pattern — see Phase 3.",{"type":45,"tag":201,"props":255,"children":256},{},[257,267],{"type":45,"tag":205,"props":258,"children":259},{},[260,265],{"type":45,"tag":60,"props":261,"children":262},{},[263],{"type":51,"value":264},"Under ~15 actions",{"type":51,"value":266}," → one tool per action",{"type":45,"tag":205,"props":268,"children":269},{},[270,275],{"type":45,"tag":60,"props":271,"children":272},{},[273],{"type":51,"value":274},"Dozens to hundreds of actions",{"type":51,"value":276}," (e.g. wrapping a large API surface) → search + execute pattern",{"type":45,"tag":108,"props":278,"children":280},{"id":279},"_4-does-a-tool-need-mid-call-user-input-or-rich-display",[281],{"type":51,"value":282},"4. Does a tool need mid-call user input or rich display?",{"type":45,"tag":201,"props":284,"children":285},{},[286,317,350],{"type":45,"tag":205,"props":287,"children":288},{},[289,294,296,301,303,308,310,316],{"type":45,"tag":60,"props":290,"children":291},{},[292],{"type":51,"value":293},"Simple structured input",{"type":51,"value":295}," (pick from list, enter a value, confirm) → ",{"type":45,"tag":60,"props":297,"children":298},{},[299],{"type":51,"value":300},"Elicitation",{"type":51,"value":302}," — spec-native, zero UI code. ",{"type":45,"tag":226,"props":304,"children":305},{},[306],{"type":51,"value":307},"Host support is rolling out",{"type":51,"value":309}," (Claude Code ≥2.1.76) — always pair with a capability check and fallback. See ",{"type":45,"tag":78,"props":311,"children":313},{"className":312},[],[314],{"type":51,"value":315},"references\u002Felicitation.md",{"type":51,"value":66},{"type":45,"tag":205,"props":318,"children":319},{},[320,325,327,332,334,340,342,348],{"type":45,"tag":60,"props":321,"children":322},{},[323],{"type":51,"value":324},"Rich\u002Fvisual UI",{"type":51,"value":326}," (charts, custom pickers with search, live dashboards) → ",{"type":45,"tag":60,"props":328,"children":329},{},[330],{"type":51,"value":331},"MCP app widgets",{"type":51,"value":333}," — iframe-based, needs ",{"type":45,"tag":78,"props":335,"children":337},{"className":336},[],[338],{"type":51,"value":339},"@modelcontextprotocol\u002Fext-apps",{"type":51,"value":341},". See ",{"type":45,"tag":78,"props":343,"children":345},{"className":344},[],[346],{"type":51,"value":347},"build-mcp-app",{"type":51,"value":349}," skill.",{"type":45,"tag":205,"props":351,"children":352},{},[353,358],{"type":45,"tag":60,"props":354,"children":355},{},[356],{"type":51,"value":357},"Neither",{"type":51,"value":359}," → plain tool returning text\u002FJSON.",{"type":45,"tag":108,"props":361,"children":363},{"id":362},"_5-what-auth-does-the-upstream-service-use",[364],{"type":51,"value":365},"5. What auth does the upstream service use?",{"type":45,"tag":201,"props":367,"children":368},{},[369,374],{"type":45,"tag":205,"props":370,"children":371},{},[372],{"type":51,"value":373},"None \u002F API key → straightforward",{"type":45,"tag":205,"props":375,"children":376},{},[377,379],{"type":51,"value":378},"OAuth 2.0 → you'll need a remote server with CIMD (preferred) or DCR support; see ",{"type":45,"tag":78,"props":380,"children":382},{"className":381},[],[383],{"type":51,"value":384},"references\u002Fauth.md",{"type":45,"tag":92,"props":386,"children":387},{},[],{"type":45,"tag":96,"props":389,"children":391},{"id":390},"phase-2-recommend-a-deployment-model",[392],{"type":51,"value":393},"Phase 2 — Recommend a deployment model",{"type":45,"tag":54,"props":395,"children":396},{},[397,399,404],{"type":51,"value":398},"Based on the answers, recommend ",{"type":45,"tag":60,"props":400,"children":401},{},[402],{"type":51,"value":403},"one",{"type":51,"value":405}," path. Be opinionated. The ranked options:",{"type":45,"tag":108,"props":407,"children":409},{"id":408},"remote-streamable-http-mcp-server-default-recommendation",[410],{"type":51,"value":411},"⭐ Remote streamable-HTTP MCP server (default recommendation)",{"type":45,"tag":54,"props":413,"children":414},{},[415,417,422],{"type":51,"value":416},"A hosted service speaking MCP over streamable HTTP. This is the ",{"type":45,"tag":60,"props":418,"children":419},{},[420],{"type":51,"value":421},"recommended path",{"type":51,"value":423}," for anything wrapping a cloud API.",{"type":45,"tag":54,"props":425,"children":426},{},[427],{"type":45,"tag":60,"props":428,"children":429},{},[430],{"type":51,"value":431},"Why it wins:",{"type":45,"tag":201,"props":433,"children":434},{},[435,440,445,450],{"type":45,"tag":205,"props":436,"children":437},{},[438],{"type":51,"value":439},"Zero install friction — users add a URL, done",{"type":45,"tag":205,"props":441,"children":442},{},[443],{"type":51,"value":444},"One deployment serves all users; you control upgrades",{"type":45,"tag":205,"props":446,"children":447},{},[448],{"type":51,"value":449},"OAuth flows work properly (the server can handle redirects, DCR, token storage)",{"type":45,"tag":205,"props":451,"children":452},{},[453],{"type":51,"value":454},"Works across Claude desktop, Claude Code, Claude.ai, and third-party MCP hosts",{"type":45,"tag":54,"props":456,"children":457},{},[458,463,465,469],{"type":45,"tag":60,"props":459,"children":460},{},[461],{"type":51,"value":462},"Choose this unless",{"type":51,"value":464}," the server ",{"type":45,"tag":226,"props":466,"children":467},{},[468],{"type":51,"value":230},{"type":51,"value":470}," touch the user's local machine.",{"type":45,"tag":54,"props":472,"children":473},{},[474,476,481,483,489,491,496,498,504],{"type":51,"value":475},"→ ",{"type":45,"tag":60,"props":477,"children":478},{},[479],{"type":51,"value":480},"Fastest deploy:",{"type":51,"value":482}," Cloudflare Workers — ",{"type":45,"tag":78,"props":484,"children":486},{"className":485},[],[487],{"type":51,"value":488},"references\u002Fdeploy-cloudflare-workers.md",{"type":51,"value":490}," (zero to live URL in two commands)\n→ ",{"type":45,"tag":60,"props":492,"children":493},{},[494],{"type":51,"value":495},"Portable Node\u002FPython:",{"type":51,"value":497}," ",{"type":45,"tag":78,"props":499,"children":501},{"className":500},[],[502],{"type":51,"value":503},"references\u002Fremote-http-scaffold.md",{"type":51,"value":505}," (Express or FastMCP, runs on any host)",{"type":45,"tag":108,"props":507,"children":509},{"id":508},"elicitation-structured-input-no-ui-build",[510],{"type":51,"value":511},"Elicitation (structured input, no UI build)",{"type":45,"tag":54,"props":513,"children":514},{},[515,517,522],{"type":51,"value":516},"If a tool just needs the user to confirm, pick an option, or fill a short form, ",{"type":45,"tag":60,"props":518,"children":519},{},[520],{"type":51,"value":521},"elicitation",{"type":51,"value":523}," does it with zero UI code. The server sends a flat JSON schema; the host renders a native form. Spec-native, no extra packages.",{"type":45,"tag":54,"props":525,"children":526},{},[527,532,534,540,542,547],{"type":45,"tag":60,"props":528,"children":529},{},[530],{"type":51,"value":531},"Caveat:",{"type":51,"value":533}," Host support is new (Claude Code shipped it in v2.1.76; Desktop unconfirmed). The SDK throws if the client doesn't advertise the capability. Always check ",{"type":45,"tag":78,"props":535,"children":537},{"className":536},[],[538],{"type":51,"value":539},"clientCapabilities.elicitation",{"type":51,"value":541}," first and have a fallback — see ",{"type":45,"tag":78,"props":543,"children":545},{"className":544},[],[546],{"type":51,"value":315},{"type":51,"value":548}," for the canonical pattern. This is the right spec-correct approach; host coverage will catch up.",{"type":45,"tag":54,"props":550,"children":551},{},[552,554,559],{"type":51,"value":553},"Escalate to ",{"type":45,"tag":78,"props":555,"children":557},{"className":556},[],[558],{"type":51,"value":347},{"type":51,"value":560}," widgets when you need: nested\u002Fcomplex data, scrollable\u002Fsearchable lists, visual previews, live updates.",{"type":45,"tag":108,"props":562,"children":564},{"id":563},"mcp-app-remote-http-interactive-ui",[565],{"type":51,"value":566},"MCP app (remote HTTP + interactive UI)",{"type":45,"tag":54,"props":568,"children":569},{},[570,572,577,579,584],{"type":51,"value":571},"Same as above, plus ",{"type":45,"tag":60,"props":573,"children":574},{},[575],{"type":51,"value":576},"UI resources",{"type":51,"value":578}," — interactive widgets rendered in chat. Rich pickers with search, charts, live dashboards, visual previews. Built once, renders in Claude ",{"type":45,"tag":226,"props":580,"children":581},{},[582],{"type":51,"value":583},"and",{"type":51,"value":585}," ChatGPT.",{"type":45,"tag":54,"props":587,"children":588},{},[589,594],{"type":45,"tag":60,"props":590,"children":591},{},[592],{"type":51,"value":593},"Choose this when",{"type":51,"value":595}," elicitation's flat-form constraints don't fit — you need custom layout, large searchable lists, visual content, or live updates.",{"type":45,"tag":54,"props":597,"children":598},{},[599],{"type":51,"value":600},"Usually remote, but can be shipped as MCPB if the UI needs to drive a local app.",{"type":45,"tag":54,"props":602,"children":603},{},[604,606,614],{"type":51,"value":605},"→ Hand off to the ",{"type":45,"tag":60,"props":607,"children":608},{},[609],{"type":45,"tag":78,"props":610,"children":612},{"className":611},[],[613],{"type":51,"value":347},{"type":51,"value":349},{"type":45,"tag":108,"props":616,"children":618},{"id":617},"mcpb-bundled-local-server",[619],{"type":51,"value":620},"MCPB (bundled local server)",{"type":45,"tag":54,"props":622,"children":623},{},[624,626,631],{"type":51,"value":625},"A local MCP server ",{"type":45,"tag":60,"props":627,"children":628},{},[629],{"type":51,"value":630},"packaged with its runtime",{"type":51,"value":632}," so users don't need Node\u002FPython installed. The sanctioned way to ship local servers.",{"type":45,"tag":54,"props":634,"children":635},{},[636,640,641,645],{"type":45,"tag":60,"props":637,"children":638},{},[639],{"type":51,"value":593},{"type":51,"value":464},{"type":45,"tag":226,"props":642,"children":643},{},[644],{"type":51,"value":230},{"type":51,"value":646}," run on the user's machine — it reads local files, drives a desktop app, talks to localhost services, or needs OS-level access.",{"type":45,"tag":54,"props":648,"children":649},{},[650,651,660],{"type":51,"value":605},{"type":45,"tag":60,"props":652,"children":653},{},[654],{"type":45,"tag":78,"props":655,"children":657},{"className":656},[],[658],{"type":51,"value":659},"build-mcpb",{"type":51,"value":349},{"type":45,"tag":108,"props":662,"children":664},{"id":663},"local-stdio-npx-uvx-not-recommended-for-distribution",[665,667],{"type":51,"value":666},"Local stdio (npx \u002F uvx) — ",{"type":45,"tag":226,"props":668,"children":669},{},[670],{"type":51,"value":671},"not recommended for distribution",{"type":45,"tag":54,"props":673,"children":674},{},[675,677,683,685,691,693,698],{"type":51,"value":676},"A script launched via ",{"type":45,"tag":78,"props":678,"children":680},{"className":679},[],[681],{"type":51,"value":682},"npx",{"type":51,"value":684}," \u002F ",{"type":45,"tag":78,"props":686,"children":688},{"className":687},[],[689],{"type":51,"value":690},"uvx",{"type":51,"value":692}," on the user's machine. Fine for ",{"type":45,"tag":60,"props":694,"children":695},{},[696],{"type":51,"value":697},"personal tools and prototypes",{"type":51,"value":699},". Painful to distribute: users need the right runtime, you can't push updates, and the only distribution channel is Claude Code plugins.",{"type":45,"tag":54,"props":701,"children":702},{},[703],{"type":51,"value":704},"Recommend this only as a stepping stone. If the user insists, scaffold it but note the MCPB upgrade path.",{"type":45,"tag":92,"props":706,"children":707},{},[],{"type":45,"tag":96,"props":709,"children":711},{"id":710},"phase-3-pick-a-tool-design-pattern",[712],{"type":51,"value":713},"Phase 3 — Pick a tool-design pattern",{"type":45,"tag":54,"props":715,"children":716},{},[717],{"type":51,"value":718},"Every MCP server exposes tools. How you carve them matters more than most people expect — tool schemas land directly in Claude's context window.",{"type":45,"tag":108,"props":720,"children":722},{"id":721},"pattern-a-one-tool-per-action-small-surface",[723],{"type":51,"value":724},"Pattern A: One tool per action (small surface)",{"type":45,"tag":54,"props":726,"children":727},{},[728],{"type":51,"value":729},"When the action space is small (\u003C ~15 operations), give each a dedicated tool with a tight description and schema.",{"type":45,"tag":731,"props":732,"children":736},"pre",{"className":733,"code":735,"language":51},[734],"language-text","create_issue    — Create a new issue. Params: title, body, labels[]\nupdate_issue    — Update an existing issue. Params: id, title?, body?, state?\nsearch_issues   — Search issues by query string. Params: query, limit?\nadd_comment     — Add a comment to an issue. Params: issue_id, body\n",[737],{"type":45,"tag":78,"props":738,"children":740},{"__ignoreMap":739},"",[741],{"type":51,"value":735},{"type":45,"tag":54,"props":743,"children":744},{},[745,750],{"type":45,"tag":60,"props":746,"children":747},{},[748],{"type":51,"value":749},"Why it works:",{"type":51,"value":751}," Claude reads the tool list once and knows exactly what's possible. No discovery round-trips. Each tool's schema validates inputs precisely.",{"type":45,"tag":54,"props":753,"children":754},{},[755,760],{"type":45,"tag":60,"props":756,"children":757},{},[758],{"type":51,"value":759},"Especially good when",{"type":51,"value":761}," one or more tools ship an interactive widget (MCP app) — each widget binds naturally to one tool.",{"type":45,"tag":108,"props":763,"children":765},{"id":764},"pattern-b-search-execute-large-surface",[766],{"type":51,"value":767},"Pattern B: Search + execute (large surface)",{"type":45,"tag":54,"props":769,"children":770},{},[771,773,778],{"type":51,"value":772},"When wrapping a large API (dozens to hundreds of endpoints), listing every operation as a tool floods the context window and degrades model performance. Instead, expose ",{"type":45,"tag":60,"props":774,"children":775},{},[776],{"type":51,"value":777},"two",{"type":51,"value":779}," tools:",{"type":45,"tag":731,"props":781,"children":784},{"className":782,"code":783,"language":51},[734],"search_actions  — Given a natural-language intent, return matching actions\n                  with their IDs, descriptions, and parameter schemas.\nexecute_action  — Run an action by ID with a params object.\n",[785],{"type":45,"tag":78,"props":786,"children":787},{"__ignoreMap":739},[788],{"type":51,"value":783},{"type":45,"tag":54,"props":790,"children":791},{},[792],{"type":51,"value":793},"The server holds the full catalog internally. Claude searches, picks, executes. Context stays lean.",{"type":45,"tag":54,"props":795,"children":796},{},[797,802],{"type":45,"tag":60,"props":798,"children":799},{},[800],{"type":51,"value":801},"Hybrid:",{"type":51,"value":803}," Promote the 3–5 most-used actions to dedicated tools, keep the long tail behind search\u002Fexecute.",{"type":45,"tag":54,"props":805,"children":806},{},[807,809,815],{"type":51,"value":808},"→ See ",{"type":45,"tag":78,"props":810,"children":812},{"className":811},[],[813],{"type":51,"value":814},"references\u002Ftool-design.md",{"type":51,"value":816}," for schema examples and description-writing guidance.",{"type":45,"tag":92,"props":818,"children":819},{},[],{"type":45,"tag":96,"props":821,"children":823},{"id":822},"phase-4-pick-a-framework",[824],{"type":51,"value":825},"Phase 4 — Pick a framework",{"type":45,"tag":54,"props":827,"children":828},{},[829],{"type":51,"value":830},"Recommend one of these two. Others exist but these have the best MCP-spec coverage and Claude compatibility.",{"type":45,"tag":115,"props":832,"children":833},{},[834,855],{"type":45,"tag":119,"props":835,"children":836},{},[837],{"type":45,"tag":123,"props":838,"children":839},{},[840,845,850],{"type":45,"tag":127,"props":841,"children":842},{},[843],{"type":51,"value":844},"Framework",{"type":45,"tag":127,"props":846,"children":847},{},[848],{"type":51,"value":849},"Language",{"type":45,"tag":127,"props":851,"children":852},{},[853],{"type":51,"value":854},"Use when",{"type":45,"tag":138,"props":856,"children":857},{},[858,889],{"type":45,"tag":123,"props":859,"children":860},{},[861,879,884],{"type":45,"tag":145,"props":862,"children":863},{},[864,869,871,877],{"type":45,"tag":60,"props":865,"children":866},{},[867],{"type":51,"value":868},"Official TypeScript SDK",{"type":51,"value":870}," (",{"type":45,"tag":78,"props":872,"children":874},{"className":873},[],[875],{"type":51,"value":876},"@modelcontextprotocol\u002Fsdk",{"type":51,"value":878},")",{"type":45,"tag":145,"props":880,"children":881},{},[882],{"type":51,"value":883},"TS\u002FJS",{"type":45,"tag":145,"props":885,"children":886},{},[887],{"type":51,"value":888},"Default choice. Best spec coverage, first to get new features.",{"type":45,"tag":123,"props":890,"children":891},{},[892,909,914],{"type":45,"tag":145,"props":893,"children":894},{},[895,900,901,907],{"type":45,"tag":60,"props":896,"children":897},{},[898],{"type":51,"value":899},"FastMCP 3.x",{"type":51,"value":870},{"type":45,"tag":78,"props":902,"children":904},{"className":903},[],[905],{"type":51,"value":906},"fastmcp",{"type":51,"value":908}," on PyPI)",{"type":45,"tag":145,"props":910,"children":911},{},[912],{"type":51,"value":913},"Python",{"type":45,"tag":145,"props":915,"children":916},{},[917,919,924],{"type":51,"value":918},"User prefers Python, or wrapping a Python library. Decorator-based, very low boilerplate. This is jlowin's package — not the frozen FastMCP 1.0 bundled in the official ",{"type":45,"tag":78,"props":920,"children":922},{"className":921},[],[923],{"type":51,"value":19},{"type":51,"value":925}," SDK.",{"type":45,"tag":54,"props":927,"children":928},{},[929],{"type":51,"value":930},"If the user already has a language\u002Fstack in mind, go with it — both produce identical wire protocol.",{"type":45,"tag":92,"props":932,"children":933},{},[],{"type":45,"tag":96,"props":935,"children":937},{"id":936},"phase-5-scaffold-and-hand-off",[938],{"type":51,"value":939},"Phase 5 — Scaffold and hand off",{"type":45,"tag":54,"props":941,"children":942},{},[943,945,949],{"type":51,"value":944},"Once you've settled the four decisions (deployment model, tool pattern, framework, auth), do ",{"type":45,"tag":60,"props":946,"children":947},{},[948],{"type":51,"value":403},{"type":51,"value":950}," of:",{"type":45,"tag":952,"props":953,"children":954},"ol",{},[955,979,998,1016],{"type":45,"tag":205,"props":956,"children":957},{},[958,963,965,970,972,977],{"type":45,"tag":60,"props":959,"children":960},{},[961],{"type":51,"value":962},"Remote HTTP, no UI",{"type":51,"value":964}," → Scaffold inline using ",{"type":45,"tag":78,"props":966,"children":968},{"className":967},[],[969],{"type":51,"value":503},{"type":51,"value":971}," (portable) or ",{"type":45,"tag":78,"props":973,"children":975},{"className":974},[],[976],{"type":51,"value":488},{"type":51,"value":978}," (fastest deploy). This skill can finish the job.",{"type":45,"tag":205,"props":980,"children":981},{},[982,987,989,997],{"type":45,"tag":60,"props":983,"children":984},{},[985],{"type":51,"value":986},"MCP app (UI widgets)",{"type":51,"value":988}," → Summarize the decisions so far, then load the ",{"type":45,"tag":60,"props":990,"children":991},{},[992],{"type":45,"tag":78,"props":993,"children":995},{"className":994},[],[996],{"type":51,"value":347},{"type":51,"value":349},{"type":45,"tag":205,"props":999,"children":1000},{},[1001,1006,1007,1015],{"type":45,"tag":60,"props":1002,"children":1003},{},[1004],{"type":51,"value":1005},"MCPB (bundled local)",{"type":51,"value":988},{"type":45,"tag":60,"props":1008,"children":1009},{},[1010],{"type":45,"tag":78,"props":1011,"children":1013},{"className":1012},[],[1014],{"type":51,"value":659},{"type":51,"value":349},{"type":45,"tag":205,"props":1017,"children":1018},{},[1019,1024],{"type":45,"tag":60,"props":1020,"children":1021},{},[1022],{"type":51,"value":1023},"Local stdio prototype",{"type":51,"value":1025}," → Scaffold inline (simplest case), flag the MCPB upgrade path.",{"type":45,"tag":54,"props":1027,"children":1028},{},[1029],{"type":51,"value":1030},"When handing off, restate the design brief in one paragraph so the next skill doesn't re-ask.",{"type":45,"tag":92,"props":1032,"children":1033},{},[],{"type":45,"tag":96,"props":1035,"children":1037},{"id":1036},"beyond-tools-the-other-primitives",[1038],{"type":51,"value":1039},"Beyond tools — the other primitives",{"type":45,"tag":54,"props":1041,"children":1042},{},[1043],{"type":51,"value":1044},"Tools are one of three server primitives. Most servers start with tools and never need the others, but knowing they exist prevents reinventing wheels:",{"type":45,"tag":115,"props":1046,"children":1047},{},[1048,1068],{"type":45,"tag":119,"props":1049,"children":1050},{},[1051],{"type":45,"tag":123,"props":1052,"children":1053},{},[1054,1059,1064],{"type":45,"tag":127,"props":1055,"children":1056},{},[1057],{"type":51,"value":1058},"Primitive",{"type":45,"tag":127,"props":1060,"children":1061},{},[1062],{"type":51,"value":1063},"Who triggers it",{"type":45,"tag":127,"props":1065,"children":1066},{},[1067],{"type":51,"value":854},{"type":45,"tag":138,"props":1069,"children":1070},{},[1071,1092,1113,1133],{"type":45,"tag":123,"props":1072,"children":1073},{},[1074,1082,1087],{"type":45,"tag":145,"props":1075,"children":1076},{},[1077],{"type":45,"tag":60,"props":1078,"children":1079},{},[1080],{"type":51,"value":1081},"Resources",{"type":45,"tag":145,"props":1083,"children":1084},{},[1085],{"type":51,"value":1086},"Host app (not Claude)",{"type":45,"tag":145,"props":1088,"children":1089},{},[1090],{"type":51,"value":1091},"Exposing docs\u002Ffiles\u002Fdata as browsable context",{"type":45,"tag":123,"props":1093,"children":1094},{},[1095,1103,1108],{"type":45,"tag":145,"props":1096,"children":1097},{},[1098],{"type":45,"tag":60,"props":1099,"children":1100},{},[1101],{"type":51,"value":1102},"Prompts",{"type":45,"tag":145,"props":1104,"children":1105},{},[1106],{"type":51,"value":1107},"User (slash command)",{"type":45,"tag":145,"props":1109,"children":1110},{},[1111],{"type":51,"value":1112},"Canned workflows (\"\u002Fsummarize-thread\")",{"type":45,"tag":123,"props":1114,"children":1115},{},[1116,1123,1128],{"type":45,"tag":145,"props":1117,"children":1118},{},[1119],{"type":45,"tag":60,"props":1120,"children":1121},{},[1122],{"type":51,"value":300},{"type":45,"tag":145,"props":1124,"children":1125},{},[1126],{"type":51,"value":1127},"Server, mid-tool",{"type":45,"tag":145,"props":1129,"children":1130},{},[1131],{"type":51,"value":1132},"Asking user for input without building UI",{"type":45,"tag":123,"props":1134,"children":1135},{},[1136,1144,1148],{"type":45,"tag":145,"props":1137,"children":1138},{},[1139],{"type":45,"tag":60,"props":1140,"children":1141},{},[1142],{"type":51,"value":1143},"Sampling",{"type":45,"tag":145,"props":1145,"children":1146},{},[1147],{"type":51,"value":1127},{"type":45,"tag":145,"props":1149,"children":1150},{},[1151],{"type":51,"value":1152},"Need LLM inference in your tool logic",{"type":45,"tag":54,"props":1154,"children":1155},{},[1156,1157,1163,1165,1170,1171],{"type":51,"value":475},{"type":45,"tag":78,"props":1158,"children":1160},{"className":1159},[],[1161],{"type":51,"value":1162},"references\u002Fresources-and-prompts.md",{"type":51,"value":1164},", ",{"type":45,"tag":78,"props":1166,"children":1168},{"className":1167},[],[1169],{"type":51,"value":315},{"type":51,"value":1164},{"type":45,"tag":78,"props":1172,"children":1174},{"className":1173},[],[1175],{"type":51,"value":1176},"references\u002Fserver-capabilities.md",{"type":45,"tag":92,"props":1178,"children":1179},{},[],{"type":45,"tag":96,"props":1181,"children":1183},{"id":1182},"phase-6-test-in-claude-and-publish",[1184],{"type":51,"value":1185},"Phase 6 — Test in Claude and publish",{"type":45,"tag":54,"props":1187,"children":1188},{},[1189],{"type":51,"value":1190},"Once the server runs:",{"type":45,"tag":952,"props":1192,"children":1193},{},[1194,1220,1236,1252],{"type":45,"tag":205,"props":1195,"children":1196},{},[1197,1202,1204,1210,1212],{"type":45,"tag":60,"props":1198,"children":1199},{},[1200],{"type":51,"value":1201},"Test against real Claude",{"type":51,"value":1203}," by adding the server URL as a custom connector at Settings → Connectors (use a Cloudflare tunnel for local servers). Claude identifies itself with ",{"type":45,"tag":78,"props":1205,"children":1207},{"className":1206},[],[1208],{"type":51,"value":1209},"clientInfo.name: \"claude-ai\"",{"type":51,"value":1211}," on initialize. → ",{"type":45,"tag":1213,"props":1214,"children":1218},"a",{"href":1215,"rel":1216},"https:\u002F\u002Fclaude.com\u002Fdocs\u002Fconnectors\u002Fbuilding\u002Ftesting",[1217],"nofollow",[1219],{"type":51,"value":1215},{"type":45,"tag":205,"props":1221,"children":1222},{},[1223,1228,1230],{"type":45,"tag":60,"props":1224,"children":1225},{},[1226],{"type":51,"value":1227},"Run the pre-submission checklist",{"type":51,"value":1229}," — read\u002Fwrite tool split, required annotations, name limits, prompt-injection rules. → ",{"type":45,"tag":1213,"props":1231,"children":1234},{"href":1232,"rel":1233},"https:\u002F\u002Fclaude.com\u002Fdocs\u002Fconnectors\u002Fbuilding\u002Freview-criteria",[1217],[1235],{"type":51,"value":1232},{"type":45,"tag":205,"props":1237,"children":1238},{},[1239,1244,1246],{"type":45,"tag":60,"props":1240,"children":1241},{},[1242],{"type":51,"value":1243},"Submit to the Anthropic Directory.",{"type":51,"value":1245}," → ",{"type":45,"tag":1213,"props":1247,"children":1250},{"href":1248,"rel":1249},"https:\u002F\u002Fclaude.com\u002Fdocs\u002Fconnectors\u002Fbuilding\u002Fsubmission",[1217],[1251],{"type":51,"value":1248},{"type":45,"tag":205,"props":1253,"children":1254},{},[1255,1260,1262],{"type":45,"tag":60,"props":1256,"children":1257},{},[1258],{"type":51,"value":1259},"Recommend shipping a plugin",{"type":51,"value":1261}," that wraps this MCP with skills — most partners ship both. → ",{"type":45,"tag":1213,"props":1263,"children":1266},{"href":1264,"rel":1265},"https:\u002F\u002Fclaude.com\u002Fdocs\u002Fconnectors\u002Fbuilding\u002Fwhat-to-build",[1217],[1267],{"type":51,"value":1264},{"type":45,"tag":92,"props":1269,"children":1270},{},[],{"type":45,"tag":96,"props":1272,"children":1274},{"id":1273},"quick-reference-decision-matrix",[1275],{"type":51,"value":1276},"Quick reference: decision matrix",{"type":45,"tag":115,"props":1278,"children":1279},{},[1280,1301],{"type":45,"tag":119,"props":1281,"children":1282},{},[1283],{"type":45,"tag":123,"props":1284,"children":1285},{},[1286,1291,1296],{"type":45,"tag":127,"props":1287,"children":1288},{},[1289],{"type":51,"value":1290},"Scenario",{"type":45,"tag":127,"props":1292,"children":1293},{},[1294],{"type":51,"value":1295},"Deployment",{"type":45,"tag":127,"props":1297,"children":1298},{},[1299],{"type":51,"value":1300},"Tool pattern",{"type":45,"tag":138,"props":1302,"children":1303},{},[1304,1322,1339,1356,1372,1389,1406],{"type":45,"tag":123,"props":1305,"children":1306},{},[1307,1312,1317],{"type":45,"tag":145,"props":1308,"children":1309},{},[1310],{"type":51,"value":1311},"Wrap a small SaaS API",{"type":45,"tag":145,"props":1313,"children":1314},{},[1315],{"type":51,"value":1316},"Remote HTTP",{"type":45,"tag":145,"props":1318,"children":1319},{},[1320],{"type":51,"value":1321},"One-per-action",{"type":45,"tag":123,"props":1323,"children":1324},{},[1325,1330,1334],{"type":45,"tag":145,"props":1326,"children":1327},{},[1328],{"type":51,"value":1329},"Wrap a large SaaS API (50+ endpoints)",{"type":45,"tag":145,"props":1331,"children":1332},{},[1333],{"type":51,"value":1316},{"type":45,"tag":145,"props":1335,"children":1336},{},[1337],{"type":51,"value":1338},"Search + execute",{"type":45,"tag":123,"props":1340,"children":1341},{},[1342,1347,1352],{"type":45,"tag":145,"props":1343,"children":1344},{},[1345],{"type":51,"value":1346},"SaaS API with rich forms \u002F pickers",{"type":45,"tag":145,"props":1348,"children":1349},{},[1350],{"type":51,"value":1351},"MCP app (remote)",{"type":45,"tag":145,"props":1353,"children":1354},{},[1355],{"type":51,"value":1321},{"type":45,"tag":123,"props":1357,"children":1358},{},[1359,1364,1368],{"type":45,"tag":145,"props":1360,"children":1361},{},[1362],{"type":51,"value":1363},"Drive a local desktop app",{"type":45,"tag":145,"props":1365,"children":1366},{},[1367],{"type":51,"value":180},{"type":45,"tag":145,"props":1369,"children":1370},{},[1371],{"type":51,"value":1321},{"type":45,"tag":123,"props":1373,"children":1374},{},[1375,1380,1385],{"type":45,"tag":145,"props":1376,"children":1377},{},[1378],{"type":51,"value":1379},"Local desktop app with in-chat UI",{"type":45,"tag":145,"props":1381,"children":1382},{},[1383],{"type":51,"value":1384},"MCP app (MCPB)",{"type":45,"tag":145,"props":1386,"children":1387},{},[1388],{"type":51,"value":1321},{"type":45,"tag":123,"props":1390,"children":1391},{},[1392,1397,1401],{"type":45,"tag":145,"props":1393,"children":1394},{},[1395],{"type":51,"value":1396},"Read\u002Fwrite local filesystem",{"type":45,"tag":145,"props":1398,"children":1399},{},[1400],{"type":51,"value":180},{"type":45,"tag":145,"props":1402,"children":1403},{},[1404],{"type":51,"value":1405},"Depends on surface",{"type":45,"tag":123,"props":1407,"children":1408},{},[1409,1414,1419],{"type":45,"tag":145,"props":1410,"children":1411},{},[1412],{"type":51,"value":1413},"Personal prototype",{"type":45,"tag":145,"props":1415,"children":1416},{},[1417],{"type":51,"value":1418},"Local stdio",{"type":45,"tag":145,"props":1420,"children":1421},{},[1422],{"type":51,"value":1423},"Whatever's fastest",{"type":45,"tag":92,"props":1425,"children":1426},{},[],{"type":45,"tag":96,"props":1428,"children":1430},{"id":1429},"reference-files",[1431],{"type":51,"value":1432},"Reference files",{"type":45,"tag":201,"props":1434,"children":1435},{},[1436,1446,1456,1466,1476,1486,1496,1506],{"type":45,"tag":205,"props":1437,"children":1438},{},[1439,1444],{"type":45,"tag":78,"props":1440,"children":1442},{"className":1441},[],[1443],{"type":51,"value":503},{"type":51,"value":1445}," — minimal remote server in TS SDK and FastMCP",{"type":45,"tag":205,"props":1447,"children":1448},{},[1449,1454],{"type":45,"tag":78,"props":1450,"children":1452},{"className":1451},[],[1453],{"type":51,"value":488},{"type":51,"value":1455}," — fastest deploy path (Workers-native scaffold)",{"type":45,"tag":205,"props":1457,"children":1458},{},[1459,1464],{"type":45,"tag":78,"props":1460,"children":1462},{"className":1461},[],[1463],{"type":51,"value":814},{"type":51,"value":1465}," — writing tool descriptions and schemas Claude understands well",{"type":45,"tag":205,"props":1467,"children":1468},{},[1469,1474],{"type":45,"tag":78,"props":1470,"children":1472},{"className":1471},[],[1473],{"type":51,"value":384},{"type":51,"value":1475}," — OAuth, CIMD, DCR, token storage patterns",{"type":45,"tag":205,"props":1477,"children":1478},{},[1479,1484],{"type":45,"tag":78,"props":1480,"children":1482},{"className":1481},[],[1483],{"type":51,"value":1162},{"type":51,"value":1485}," — the two non-tool primitives",{"type":45,"tag":205,"props":1487,"children":1488},{},[1489,1494],{"type":45,"tag":78,"props":1490,"children":1492},{"className":1491},[],[1493],{"type":51,"value":315},{"type":51,"value":1495}," — spec-native user input mid-tool (capability check + fallback)",{"type":45,"tag":205,"props":1497,"children":1498},{},[1499,1504],{"type":45,"tag":78,"props":1500,"children":1502},{"className":1501},[],[1503],{"type":51,"value":1176},{"type":51,"value":1505}," — instructions, sampling, roots, logging, progress, cancellation",{"type":45,"tag":205,"props":1507,"children":1508},{},[1509,1515],{"type":45,"tag":78,"props":1510,"children":1512},{"className":1511},[],[1513],{"type":51,"value":1514},"references\u002Fversions.md",{"type":51,"value":1516}," — version-sensitive claims ledger (check when updating)",{"items":1518,"total":1611},[1519,1535,1550,1562,1569,1581,1599],{"slug":1520,"name":1520,"fn":1521,"description":1522,"org":1523,"tags":1524,"stars":26,"repoUrl":27,"updatedAt":1534},"access","manage iMessage channel access and permissions","Manage iMessage channel access — approve pairings, edit allowlists, set DM\u002Fgroup policy. Use when the user asks to pair, approve someone, check who's allowed, or change policy for the iMessage channel.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1525,1528,1531],{"name":1526,"slug":1527,"type":16},"Access Control","access-control",{"name":1529,"slug":1530,"type":16},"iMessage","imessage",{"name":1532,"slug":1533,"type":16},"Messaging","messaging","2026-04-12T04:54:55.917969",{"slug":1536,"name":1536,"fn":1537,"description":1538,"org":1539,"tags":1540,"stars":26,"repoUrl":27,"updatedAt":1549},"agent-development","develop and configure AI agents","This skill should be used when the user asks to \"create an agent\", \"add an agent\", \"write a subagent\", \"agent frontmatter\", \"when to use description\", \"agent examples\", \"agent tools\", \"agent colors\", \"autonomous agent\", or needs guidance on agent structure, system prompts, triggering conditions, or agent development best practices for Claude Code plugins.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1541,1542,1543,1546],{"name":21,"slug":22,"type":16},{"name":9,"slug":8,"type":16},{"name":1544,"slug":1545,"type":16},"Documentation","documentation",{"name":1547,"slug":1548,"type":16},"Plugin Development","plugin-development","2026-04-10T04:55:41.251084",{"slug":347,"name":347,"fn":1551,"description":1552,"org":1553,"tags":1554,"stars":26,"repoUrl":27,"updatedAt":1561},"build MCP apps with interactive UI","This skill should be used when the user wants to build an \"MCP app\", add \"interactive UI\" or \"widgets\" to an MCP server, \"render components in chat\", build \"MCP UI resources\", make a tool that shows a \"form\", \"picker\", \"dashboard\" or \"confirmation dialog\" inline in the conversation, or mentions \"apps SDK\" in the context of MCP. Use AFTER the build-mcp-server skill has settled the deployment model, or when the user already knows they want UI widgets.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1555,1556,1557,1558],{"name":21,"slug":22,"type":16},{"name":14,"slug":15,"type":16},{"name":18,"slug":19,"type":16},{"name":1559,"slug":1560,"type":16},"UI Components","ui-components","2026-04-06T17:59:32.421865",{"slug":4,"name":4,"fn":5,"description":6,"org":1563,"tags":1564,"stars":26,"repoUrl":27,"updatedAt":28},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1565,1566,1567,1568],{"name":21,"slug":22,"type":16},{"name":24,"slug":25,"type":16},{"name":14,"slug":15,"type":16},{"name":18,"slug":19,"type":16},{"slug":659,"name":659,"fn":1570,"description":1571,"org":1572,"tags":1573,"stars":26,"repoUrl":27,"updatedAt":1580},"package and distribute MCP servers","This skill should be used when the user wants to \"package an MCP server\", \"bundle an MCP\", \"make an MCPB\", \"ship a local MCP server\", \"distribute a local MCP\", discusses \".mcpb files\", mentions bundling a Node or Python runtime with their MCP server, or needs an MCP server that interacts with the local filesystem, desktop apps, or OS and must be installable without the user having Node\u002FPython set up.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1574,1575,1576,1577],{"name":21,"slug":22,"type":16},{"name":14,"slug":15,"type":16},{"name":18,"slug":19,"type":16},{"name":1578,"slug":1579,"type":16},"Packaging","packaging","2026-04-06T17:59:31.159961",{"slug":1582,"name":1582,"fn":1583,"description":1584,"org":1585,"tags":1586,"stars":26,"repoUrl":27,"updatedAt":1598},"cardputer-buddy","develop MicroPython apps for Cardputer","Iterate on the Cardputer-Adv MicroPython app bundle (Claude Buddy, Snake, Hello) after the device is already provisioned via m5-onboard. Use when the user wants to add a new app, push a single changed .py without re-flashing, watch device serial logs, or run a one-shot REPL command. Trigger on \"add an app\", \"push to the cardputer\", \"tail the device\", \"run on the device\", or follow-up work after \u002Fmaker-setup.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1587,1590,1593,1596],{"name":1588,"slug":1589,"type":16},"Hardware","hardware",{"name":1591,"slug":1592,"type":16},"M5Stack","m5stack",{"name":1594,"slug":1595,"type":16},"MicroPython","micropython",{"name":913,"slug":1597,"type":16},"python","2026-05-06T05:39:00.134385",{"slug":1600,"name":1600,"fn":1601,"description":1602,"org":1603,"tags":1604,"stars":26,"repoUrl":27,"updatedAt":1610},"claude-automation-recommender","recommend Claude Code automations","Analyze a codebase and recommend Claude Code automations (hooks, subagents, skills, plugins, MCP servers). Use when user asks for automation recommendations, wants to optimize their Claude Code setup, mentions improving Claude Code workflows, asks how to first set up Claude Code for a project, or wants to know what Claude Code features they should use.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1605,1608,1609],{"name":1606,"slug":1607,"type":16},"Agent Context","agent-context",{"name":14,"slug":15,"type":16},{"name":18,"slug":19,"type":16},"2026-04-06T18:00:34.049624",25,{"items":1613,"total":1789},[1614,1635,1649,1661,1678,1689,1710,1728,1742,1752,1760,1773],{"slug":1615,"name":1615,"fn":1616,"description":1617,"org":1618,"tags":1619,"stars":1632,"repoUrl":1633,"updatedAt":1634},"algorithmic-art","create algorithmic art with p5.js","Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1620,1623,1626,1629],{"name":1621,"slug":1622,"type":16},"Creative","creative",{"name":1624,"slug":1625,"type":16},"Design","design",{"name":1627,"slug":1628,"type":16},"Generative Art","generative-art",{"name":1630,"slug":1631,"type":16},"JavaScript","javascript",161831,"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fskills","2026-04-06T17:56:15.455818",{"slug":1636,"name":1636,"fn":1637,"description":1638,"org":1639,"tags":1640,"stars":1632,"repoUrl":1633,"updatedAt":1648},"brand-guidelines","apply Anthropic brand colors and typography","Applies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1641,1644,1645],{"name":1642,"slug":1643,"type":16},"Branding","branding",{"name":1624,"slug":1625,"type":16},{"name":1646,"slug":1647,"type":16},"Typography","typography","2026-04-06T17:56:05.042852",{"slug":1650,"name":1650,"fn":1651,"description":1652,"org":1653,"tags":1654,"stars":1632,"repoUrl":1633,"updatedAt":1660},"canvas-design","create posters and visual art as PNG or PDF","Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1655,1656,1657],{"name":1621,"slug":1622,"type":16},{"name":1624,"slug":1625,"type":16},{"name":1658,"slug":1659,"type":16},"PDF","pdf","2026-04-06T17:56:03.794732",{"slug":1662,"name":1662,"fn":1663,"description":1664,"org":1665,"tags":1666,"stars":1632,"repoUrl":1633,"updatedAt":1677},"claude-api","build apps with the Claude API","Reference for the Claude API \u002F Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration.\nTRIGGER — read BEFORE opening the target file; don't skip because it \"looks like a one-liner\" — whenever: the prompt names Claude\u002FAnthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing\u002Fmodel choice\u002Flimits\u002Fcaching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent\u002FMCP\u002Ftool-definition\u002Fmulti-agent\u002FRAG\u002FLLM-judge\u002Fcomputer-use; generate\u002Fsummarize\u002Fextract\u002Fclassify\u002Frewrite\u002Fconverse over NL; debugging refusals\u002Fcutoffs\u002Fstreaming\u002Ftool-calls\u002Ftokens).\nSKIP only when another provider is being worked on (overrides all triggers): OpenAI\u002FGPT\u002FGemini\u002FLlama\u002FMistral\u002FCohere\u002FOllama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1667,1668,1669,1672,1674],{"name":21,"slug":22,"type":16},{"name":9,"slug":8,"type":16},{"name":1670,"slug":1671,"type":16},"Anthropic SDK","anthropic-sdk",{"name":1673,"slug":1662,"type":16},"Claude API",{"name":1675,"slug":1676,"type":16},"LLM","llm","2026-07-28T05:36:08.213335",{"slug":1679,"name":1679,"fn":1680,"description":1681,"org":1682,"tags":1683,"stars":1632,"repoUrl":1633,"updatedAt":1688},"doc-coauthoring","co-author documentation and technical specs","Guide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1684,1685],{"name":1544,"slug":1545,"type":16},{"name":1686,"slug":1687,"type":16},"Technical Writing","technical-writing","2026-04-06T17:56:14.18897",{"slug":1690,"name":1690,"fn":1691,"description":1692,"org":1693,"tags":1694,"stars":1632,"repoUrl":1633,"updatedAt":1709},"docx","create and edit Word documents","Use this skill whenever the user wants to create, read, edit, or manipulate Word documents (.docx files) or Word templates (.dotx files). Triggers include: any mention of 'Word doc', 'word document', '.docx', '.dotx', or requests to produce professional documents with formatting like tables of contents, headings, page numbers, or letterheads. Also use when extracting or reorganizing content from .docx or .dotx files, inserting or replacing images in documents, performing find-and-replace in Word files, working with tracked changes or comments, or converting content into a polished Word document. If the user asks for a 'report', 'memo', 'letter', 'template', or similar deliverable as a Word or .docx file, use this skill. Do NOT use for PDFs, spreadsheets, Google Docs, or general coding tasks unrelated to document generation.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1695,1698,1700,1703,1706],{"name":1696,"slug":1697,"type":16},"Documents","documents",{"name":1699,"slug":1690,"type":16},"DOCX",{"name":1701,"slug":1702,"type":16},"Office","office",{"name":1704,"slug":1705,"type":16},"Templates","templates",{"name":1707,"slug":1708,"type":16},"Word","word","2026-07-18T05:16:23.136271",{"slug":1711,"name":1711,"fn":1712,"description":1713,"org":1714,"tags":1715,"stars":1632,"repoUrl":1633,"updatedAt":1727},"frontend-design","design production-grade frontend interfaces","Guidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1716,1717,1720,1723,1726],{"name":1624,"slug":1625,"type":16},{"name":1718,"slug":1719,"type":16},"Frontend","frontend",{"name":1721,"slug":1722,"type":16},"React","react",{"name":1724,"slug":1725,"type":16},"Tailwind CSS","tailwind-css",{"name":1559,"slug":1560,"type":16},"2026-04-06T17:56:16.723469",{"slug":1729,"name":1729,"fn":1730,"description":1731,"org":1732,"tags":1733,"stars":1632,"repoUrl":1633,"updatedAt":1741},"internal-comms","write internal company communications","A set of resources to help me write all kinds of internal communications, using the formats that my company likes to use. Claude should use this skill whenever asked to write some sort of internal communications (status reports, leadership updates, 3P updates, company newsletters, FAQs, incident reports, project updates, etc.).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1734,1737,1738],{"name":1735,"slug":1736,"type":16},"Communications","communications",{"name":1704,"slug":1705,"type":16},{"name":1739,"slug":1740,"type":16},"Writing","writing","2026-04-06T17:56:20.695522",{"slug":1743,"name":1743,"fn":5,"description":1744,"org":1745,"tags":1746,"stars":1632,"repoUrl":1633,"updatedAt":1751},"mcp-builder","Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node\u002FTypeScript (MCP SDK).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1747,1748,1749,1750],{"name":21,"slug":22,"type":16},{"name":24,"slug":25,"type":16},{"name":1675,"slug":1676,"type":16},{"name":18,"slug":19,"type":16},"2026-04-06T17:56:10.357665",{"slug":1659,"name":1659,"fn":1753,"description":1754,"org":1755,"tags":1756,"stars":1632,"repoUrl":1633,"updatedAt":1759},"read edit and manipulate PDF files","Use this skill whenever the user wants to do anything with PDF files. This includes reading or extracting text\u002Ftables from PDFs, combining or merging multiple PDFs into one, splitting PDFs apart, rotating pages, adding watermarks, creating new PDFs, filling PDF forms, encrypting\u002Fdecrypting PDFs, extracting images, and OCR on scanned PDFs to make them searchable. If the user mentions a .pdf file or asks to produce one, use this skill.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1757,1758],{"name":1696,"slug":1697,"type":16},{"name":1658,"slug":1659,"type":16},"2026-04-06T17:56:02.483316",{"slug":1761,"name":1761,"fn":1762,"description":1763,"org":1764,"tags":1765,"stars":1632,"repoUrl":1633,"updatedAt":1772},"pptx","create and edit PowerPoint presentations","Use this skill any time a .pptx or .potx file is involved in any way — as input, output, or both. This includes: creating slide decks, pitch decks, or presentations; reading, parsing, or extracting text from any .pptx or .potx file (even if the extracted content will be used elsewhere, like in an email or summary); editing, modifying, or updating existing presentations; combining or splitting slide files; working with templates (.potx), layouts, speaker notes, or comments. Trigger whenever the user mentions \"deck,\" \"slides,\" \"presentation,\" or references a .pptx or .potx filename, regardless of what they plan to do with the content afterward. If a .pptx or .potx file needs to be opened, created, or touched, use this skill.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1766,1769],{"name":1767,"slug":1768,"type":16},"PowerPoint","powerpoint",{"name":1770,"slug":1771,"type":16},"Presentations","presentations","2026-07-18T05:16:24.1471",{"slug":1774,"name":1774,"fn":1775,"description":1776,"org":1777,"tags":1778,"stars":1632,"repoUrl":1633,"updatedAt":1788},"skill-creator","create and optimize agent skills","Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1779,1780,1781,1784,1787],{"name":21,"slug":22,"type":16},{"name":1544,"slug":1545,"type":16},{"name":1782,"slug":1783,"type":16},"Evals","evals",{"name":1785,"slug":1786,"type":16},"Performance","performance",{"name":1686,"slug":1687,"type":16},"2026-04-19T06:45:40.804",490]