Developer Platform / MCP

Generate presentations from your AI agent.

The Skyline Studio MCP Server exposes presentation generation as a Model Context Protocol tool surface. Connect Claude, Cursor, or any MCP-aware agent and let it design world-class decks on your behalf.

Endpoint

https://mcp.skyline-studio.ai/mcp

Start for free Connect a client

Speed

Watch it build

See your deck come together live as the agent works—real-time progress from first prompt to a finished, polished presentation.

Quality

Boardroom-ready

Sophisticated design systems ensure every deck looks polished, refined, and executive-ready.

Output

Editable decks

Generations land in your Workspace as fully structured, editable slides - ready to refine, present, and share.

Connect

One config. Any MCP-aware client.

Add the endpoint to your client's MCP configuration and complete the OAuth handshake the first time you call a tool. The server runs as your authenticated Skyline Studio user, so generated presentations land in the same workspace you use in the web app.

1Open your MCP client config (Claude, Cursor, etc.).
2Add the Skyline Studio server entry using the snippet on the right.
3Trigger a tool call. Your client opens an OAuth browser tab once; subsequent calls run on the persisted session.

mcp.json

{
  "mcpServers": {
    "skyline-studio": {
      "url": "https://mcp.skyline-studio.ai/mcp"
    }
  }
}

ClaudeConnectors

Cursor~/.cursor/mcp.json

Codex / OpenAImcpServers array

Custom agentsStreamable HTTP MCP SDK

Tools

Six tools, one creative surface.

Every tool returns a stable JSON shape and is described to your agent with rich, AI-readable schemas so it can pick the right one without hand-holding.

generate_presentation_start

Start a generation

Kicks off a new presentation generation job. Long-polls inline by default so a single tool call typically returns the finished deck. Generations usually finish in 5–7 minutes.

userPrompt: Prompt describing the target presentation.
systemInstructions: AI-to-AI guidance with three fields: designStyle, resourcesToUse, context.
slideCount: Target number of slides. 1–25, default 10.
generateImages: Whether the model should generate images.
themeId: Optional theme id from list_available_themes.
images: Optional image attachments (max 8). Each is a public imageUrl forwarded as a visual reference.
waitForCompletionMs: Long-poll budget in milliseconds. Recommended: 220000. Maximum: 900000.

generate_presentation_status

Check generation status

Returns the status of a running job. Pass waitForCompletionMs to long-poll until the job reaches a terminal state and to receive MCP progress notifications via your progressToken.

jobId: Job UUID returned by generate_presentation_start.
waitForCompletionMs: Long-poll budget in milliseconds.

generate_presentation_cancel

Cancel a generation

Cancels a running generation. No-op if the job has already reached a terminal status.

jobId: Job UUID to cancel.

list_available_themes

List curated themes

Returns the catalog of curated design themes. Each theme ships with a ready-to-use designStyle prompt string that you can pass verbatim as systemInstructions.designStyle.

get_billing_status

Read billing & AI credits

Returns billing_plan and remaining ai_credits for the authenticated user, plus the URL to share if the user needs to upgrade or top up.

health_check

Health check

Returns basic backend MCP process status. Useful for verifying connectivity.

Example

Call generate_presentation_start

A typical first call. The server enriches the request with the selected theme's reference image, queues the generation, and long-polls until a terminal status. The response includes an editorUrl to open the deck in Skyline Studio and a downloadUrl for the exported PDF.

Recommended waitForCompletionMs: 220000 (7 minutes) to avoid repeated polls.
Send a progressToken in _meta to receive incremental progress notifications.
Attach up to 8 reference images via images[].imageUrl for visual direction or data context.

tools/call · generate_presentation_start

{
  "userPrompt": "10-slide investor update for our Series A. Highlight growth, key wins this quarter, roadmap, and the ask.",
  "systemInstructions": {
    "designStyle": "Swiss editorial, restrained palette, Inter typography.",
    "resourcesToUse": "Internal KPI snapshot, recent customer logos.",
    "context": "Audience: existing investors. Tone: confident, evidence-based."
  },
  "slideCount": 10,
  "generateImages": true,
  "themeId": "swiss-editorial",
  "waitForCompletionMs": 220000
}

Authentication

OAuth, no API keys.

The server advertises an OAuth 2.0 protected resource at/.well-known/oauth-protected-resource. Compliant MCP clients discover the authorization server, complete the flow in a browser, and store the bearer for subsequent tool calls. Every request is scoped to the authenticated user and organization.

Billing

Pay with AI credits.

Generations consume AI credits from your Skyline Studio plan. If the balance hits zero,generate_presentation_startreturns status: "insufficient_credits"with an upgradeUrl your agent can share with the user. Callget_billing_status any time for a live snapshot.

Ship presentations from your agent today.

Start for free Talk to the team