MCP server
Also known as: Model Context Protocol server, MCP, agent tool server
An MCP server is a process that exposes tools, resources, and prompts to an LLM client over the Model Context Protocol — Anthropic's open standard, donated to the Linux Foundation in December 2025. MCP servers let agents in Claude, Cursor, Windsurf, and VS Code call your API as native tools.
Example
Lamina's MCP server exposes five tools — lamina.create, lamina.track, lamina.evaluate, lamina.distribute, lamina.brand_lookup — over a single npx @lamina/mcp install. Claude, Cursor, and Windsurf pick them up without further configuration.
How MCP works under the hood
MCP is a JSON-RPC 2.0 protocol. Messages flow as request/response and notification frames over stdio (for locally spawned servers) or HTTP+SSE (for remote ones). On connect, client and server negotiate capabilities. The server can expose three primitives: tools (callable functions), resources (readable data the client can attach to context), and prompts (user-selectable templates). Anthropic introduced it in November 2024; in December 2025 the spec was donated to the Agentic AI Foundation under the Linux Foundation.
Why MCP matters for media
Before MCP, every agent integration was a bespoke tool wrapper. With MCP, a media platform exposes one server and every MCP-aware client — Claude Desktop, Cursor, Windsurf, Claude Code, VS Code's GitHub Copilot Chat, ChatGPT desktop — picks it up automatically. That's why generative media for agents shipped with MCP first: it's the discovery layer the agent ecosystem has standardized on.
References
Related terms
Agent-callable API
An agent-callable API is one designed to be invoked by an autonomous LLM agent rather than a human developer. It exposes typed tool schemas, returns structured outputs an agent can parse, supports idempotency keys for safe retries, and handles long-running jobs through streaming or polling instead of blocking.
Generative media for agents
Generative media for agents is the category of image, video, audio, and 3D APIs designed to be called by autonomous AI agents — not humans clicking a UI. They expose tools through MCP, return structured outputs an LLM can parse, and tolerate retries, long jobs, and failure modes only an agent encounters.
Creative API
A Creative API is a developer interface that wraps multiple generative models, brand context, evaluation, and distribution behind a single endpoint. Instead of calling raw model endpoints — Sora, Veo, FLUX, Runway — directly, you call a creative app like productShoot or virtualTryOn and the API handles routing, evaluating, and delivering the result.
Idempotent generation
Idempotent generation is the property that the same generative-media call — same brief, brand, seed, and idempotency key — returns the same result on retry instead of producing a new (and billed) output. It's a survival requirement for agents, which retry on timeout and would otherwise create duplicate runs.