Developers · MCP Server

Plug your library straight into an AI agent

A native Model Context Protocol server lives inside the product. Eleven first-class tools let agents search, organize, upload, share, and update assets, with the same scoped tokens and the same audit trail as the rest of the API.

Try the live product Read about the REST API

What is MCP

A standard way for an agent to call external tools

The Model Context Protocol is an open standard for describing tools, resources, and prompts to an AI client in a way that a language model can call them safely. Pick an MCP-aware client, point it at the Freedam server, hand it a token, and the agent suddenly knows how to drive your library.

Because the server runs inside the product, every tool call honors the same permission model as a human user. A token with read-only abilities cannot be coerced into deleting an asset, no matter how creative the prompt.

About the MCP standard

Example client config

// Example MCP client configuration
{
  "mcpServers": {
    "freedam": {
      "url": "https://{workspace}.freedam.io/mcp/freedam",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}

Eleven tools, ready to call

Each tool maps to a normal API capability, with a schema-validated input. No private endpoints, no agent-only side doors.

search-assets

Search the library with natural language. Hybrid full-text plus semantic ranking, optional rule-based filters.

get-asset

Fetch one asset by global ID. Returns metadata, conversions, and rights.

update-asset-metadata

Edit titles, descriptions, keywords, AI fields, and structured metadata, with validation.

list-collections

Browse collections by name, parent, or only the roots. Paginated.

create-collection

Create a new collection, optionally under a parent and with a chosen type.

add-to-collection

Add a list of assets to an existing collection in one call.

upload-assets

Open an upload batch and stream files into it. Each batch can land in a target collection.

get-operation-status

Poll a long-running operation and read its progress, errors, and result counts.

create-share

Generate a share link for an asset or collection, with expiration, watermark, and access controls.

get-vocabularies

Read controlled vocabularies and their terms. Useful when an agent needs to pick valid values.

bulk-update

Apply a set of field changes to many assets at once and track the result through an operation.

Same scoped tokens

The MCP server uses the same token system as the REST API. Issue a token with the abilities the agent needs, and nothing more.

Same audit trail

Every tool call lands in the same audit log as a human action. You always know which token did what, when, and to which asset.

Same rate limits

Token rate limits apply to MCP calls too. A runaway agent loop hits the same ceiling as a runaway script.

Where it shines

A few patterns we have seen teams pick up quickly once the MCP server is connected.

Hand off curation to an agent

Let a language model triage a fresh upload batch: deduplicate, fill missing metadata, suggest collections, all through scoped tool calls.

Conversational search for non-technical teams

Wire your team’s chat assistant into the library. Marketing asks for "the spring 2025 product photos with a cream background"; the agent searches, reads, and shares.

Agentic publishing workflows

A planner agent picks assets, an editor agent writes captions, a reviewer agent applies the brand checklist, all hitting the same MCP server.

Developers · MCP

Give your agents safe access to your library

Spin up a token, point your MCP-aware client at the server, and let the model do the rest, inside the perimeter you defined.

Try the live product Read about the REST API