MCP (Model Context Protocol)

What it means

MCP servers expose capabilities (search files, query tickets, run commands) that compatible clients discover and call in a standardized way.

Why designers should care

MCP changes what agents can touch; UX must show connected sources, permissions, connection health, and what data left the user’s environment.

Example

A design copilot connects to Figma and Linear via MCP; the UI lists active integrations, scopes (“This project only”), and logs each tool call with a link to the affected issue.

Common mistakes

• • Enabling powerful MCP tools with no consent or scope UI.
• • Tool errors shown as generic “AI failed” with no integration name.
• • Assuming designers do not need to design connection onboarding and revoke flows.

Related terms

• Tool Use: The model emits structured calls (name + arguments) that your app executes, then feeds results back into the conversation for the next step.
• Harness: The harness connects the model to file systems, APIs, browsers, terminals, and review gates, enforcing limits on steps, permissions, and side effects.
• Agent: The model decides sequences of actions (search, draft, click, call APIs) within policies until the task completes or hits a human checkpoint.
• Prompt Injection: Malicious or accidental instructions embedded in retrieved or pasted content override intended rules (“ignore previous instructions and…”).
• Orchestration: An orchestrator decides what runs next: call search, pass results to the LLM, invoke a tool, branch on validation, or pause for approval.

Explore more

• Agentic UX framework