Model Context Protocol

Your content warehouse
as an MCP server.

Every account ships with its own Model Context Protocol endpoint. Claude, ChatGPT, and custom agents read your approved library directly — no SDK to install, no glue code to maintain.

Start free trial See integrations →

Primer

What is MCP?

Model Context Protocol (MCP) is an open standard — originally published by Anthropic — that defines how AI agents talk to external systems. Think of it as "the USB port for LLMs": a universal connector that any agent can plug into any compliant data source.

Before MCP, hooking an AI agent up to your data meant building a bespoke tool-use integration, handling auth, marshalling responses, and keeping the prompt template in sync with your schema. Every new agent meant repeating the work.

With MCP, you expose your data once — as a standard endpoint — and every MCP-compatible agent can read it. Clipsilo gives every account one of these endpoints, scoped to your approved content library, out of the box.

Architecture

From agent to asset in one hop

AI agent

Claude, ChatGPT, LangChain, custom tooling

MCP endpoint

mcp://clipsilo.io/mcp/{slug}

Clipsilo warehouse

Approved assets + rich metadata

Your S3 bucket

Signed URLs, stable references

The agent issues a standard MCP tool call. Clipsilo authenticates, scopes to your account's approved assets, resolves SHAs to signed S3 URLs, and returns structured results — all in under a second.

Example tool call

One call. Only approved assets.

clipsilo_search_assets

// Agent request
{
  "tool": "clipsilo_search_assets",
  "input": {
    "tags": ["campaign", "q3"],
    "status": "approved",
    "content_type": "video",
    "limit": 5
  }
}

// Response
{
  "assets": [
    {
      "id": "ast_01j9x8k3pqrs...",
      "title": "Brand Campaign Q3 — Hero Video",
      "tags": ["campaign", "q3", "hero"],
      "content_type": "video/mp4",
      "duration_seconds": 62,
      "download_url": "https://your-bucket.s3.amazonaws.com/...?X-Amz-Signature=...",
      "approved_at": "2025-07-14T10:22:00Z",
      "approved_by": "sarah@agency.com",
      "platform_assets": [
        { "platform": "youtube",   "caption": "Q3 Brand Campaign" },
        { "platform": "instagram", "caption": "Q3 · summer drop ⚡" }
      ]
    }
  ]
}

Results are pre-filtered to approved-only. The agent never sees drafts, rejected content, or assets from other accounts. Every download URL is fresh, signed, and scoped.

Use cases

What teams build with Clipsilo MCP

AI-powered social posting

A Claude agent drafts captions, picks the right approved hero video from your warehouse, and schedules posts across TikTok, Instagram, and LinkedIn — all without a human manually handing it a file.

Automated content reuse

A nightly agent scans your approved library, spots videos that haven't been reposted in 90 days, and generates fresh captions for a second run. Evergreen content stops dying in a folder.

Agent-based distribution

Give an agent a campaign brief. It pulls matching approved assets from Clipsilo, pushes them through transcoding, and delivers platform-correct variants to a dozen destinations in parallel.

Creative brief lookup

Producers ask an agent "what hero videos did we approve for the spring campaign?" — and get a ranked, linked list back in the chat, with download URLs ready to drop into the next deck.

Compatible agents

Works with every MCP-aware agent

If it speaks MCP, it speaks Clipsilo. That list is growing fast — here are the ones we test against today.

Claude (Anthropic)

Native MCP support in Claude Desktop and Claude API. Paste your MCP URL into settings and go.

ChatGPT (with plugins)

LangChain / custom agents

Drop-in MCP client support via the standard SDK. Works with any Python or TypeScript agent runtime.

Give your agents real content — approved, addressable, agent-ready.

Start your 7-day free trial. Your MCP endpoint is live the moment you sign up.

Start free trial See pricing

7-day free trial · No credit card · Cancel anytime

Your content warehouse as an MCP server.