klamdo-mcp

MCP server for Klamdo — connect your AI agent to Klamdo's creative engine to generate images, videos, and full agentic content packages.

What this does

Klamdo is a structured content generation backend. This MCP server exposes its capabilities so any MCP-compatible AI agent (Claude, GPT-4o, custom agents) can:

• Generate identity-locked 4K images and 5-second vertical videos from a single reference photo
• Create agentic packages — structured bundles of scripts, captions, images, videos, voiceovers, ebooks, and slide decks generated in one shot
• Receive machine-readable deliverables via the package manifest — a typed contract your agent uses to build landing pages, schedule posts, assemble ebooks, and distribute content downstream

Quick start

stdio (Claude Desktop, local agents)

npx klamdo-mcp

Set your API key via environment variable:

KLAMDO_API_KEY=your_api_key npx klamdo-mcp

Get your key at klamdo.app/profile.

Claude Desktop config

{
  "mcpServers": {
    "klamdo": {
      "command": "npx",
      "args": ["klamdo-mcp"],
      "env": {
        "KLAMDO_API_KEY": "your_api_key_here"
      }
    }
  }
}

HTTP (Smithery / remote agents)

Available at https://mcp.klamdo.app/mcp — connect via Smithery or any HTTP MCP client with your API key as the Authorization: Bearer header.

Tools

| Tool | Description | |---|---| | get_account | Credit balance, plan tier, free sample eligibility | | list_jobs | Recent generation jobs, newest first | | generate_image | Start a 4K image generation job (3 credits) | | generate_video | Start a 5-second vertical video job (6 credits) | | upload_start_frame | Upload a reference image for video generation | | check_job | Poll a job for status and download URLs | | list_packages | List available agentic packages with deliverable breakdown | | create_package | Start an agentic package job from a business prompt | | get_package | Poll a package for status and progress | | get_package_manifest | Fetch the structured deliverables manifest for downstream agent use |

Image / video workflow

1. upload_start_frame({ imageUrl: "https://your-cdn.com/photo.jpg" })
   → { assetId: "asset_xxx" }

2. generate_video({ prompt: "...", startFrameAssetId: "asset_xxx" })
   → { jobId: "job_xxx" }

3. check_job({ jobId: "job_xxx" })
   → { status: "completed", assets: [{ kind: "video", url: "..." }] }

Agentic package workflow

Packages generate a full content bundle in one call — scripts, images, videos, voiceover, ebook outline, slide deck, captions — and return a typed manifest your agent can use to build and distribute.

1. list_packages({ accessibleOnly: true })
   → see what's available on your plan

2. create_package({
     packageType: "coaching-brand",
     prompt: "Launch package for my 90-day coaching offer targeting new entrepreneurs",
     niche: "business coaching",
     avatarDurationSeconds: 60
   })
   → { id: "pkg_xxx", manifestUrl: "/api/mcp/packages/pkg_xxx/manifest" }

3. get_package({ packageId: "pkg_xxx" })
   → { status: "processing", progress: { completed: 4, total: 11 } }

4. get_package_manifest({ packageId: "pkg_xxx" })
   → structured manifest with text[], media[], documents[], buildHints

Package manifest

The manifest is a machine-readable contract — not just a list of files. Each deliverable has a typed role so your agent knows what to do with it:

{
  "outputs": {
    "text": [
      { "role": "script", "text": "..." },
      { "role": "landing_page_copy", "text": "..." },
      { "role": "email_sequence", "text": "..." }
    ],
    "media": [
      { "role": "hero_image", "url": "https://..." },
      { "role": "avatar_video", "url": "https://..." }
    ],
    "documents": [
      { "role": "slide_deck", "url": "https://..." },
      { "role": "pdf_ebook", "url": "https://..." }
    ]
  },
  "buildHints": {
    "primaryOffer": "...",
    "primaryCTA": "...",
    "recommendedExecutionOrder": ["landing_page_copy", "hero_image", "avatar_video", ...],
    "heroAssetIds": ["..."],
    "landingPageSourceIds": ["..."]
  }
}

Your agent fans out: builds the landing page from landing_page_copy, populates the ebook from ebook_outline, schedules the avatar_video, posts the hero_image — all from a single manifest.

Package types

| Package | Min tier | Key outputs | |---|---|---| | digital-marketing | Starter | scripts, images, video, avatar video, voiceover | | social-media | Starter | captions, content calendar, images, avatar video | | ugc-ad | Creator | UGC scripts, ad images, ad video | | coaching-brand | Creator | scripts, avatar video, slide deck, PDF lead magnet | | product-launch | Pro | landing page copy, email sequence, images, avatar video, slide deck | | content-repurpose | Starter | repurposed text, images, video, slide deck | | motion-control | Creator | motion-controlled video | | faceless-content | Creator | faceless video with voiceover | | build-your-own-ai | Starter | 10 AI character images (realistic/anime/cartoon/fantasy/cyber/3D) |

Auth

All tools require a Klamdo API key. Get yours at klamdo.app/profile.

• stdio: set KLAMDO_API_KEY environment variable
• HTTP: pass Authorization: Bearer <key> header

Requirements

• Node.js ≥ 18
• A Klamdo account with an agentic plan tier for package tools (free sample available for image/video)

Issues

github.com/klamdo-app/klamdo-mcp/issues