MCP enables AI agents to access APIs with granular data controls

MCP is introducing new methods for enabling AI agents to interact with company APIs while emphasizing data security. Instead of exposing all information to AI agents, engineers can now use field-level contracts—often via GraphQL—which precisely control what fields agents can access or modify. This approach allows organizations to tailor agent permissions to different teams without building numerous overlapping tools, reducing the risk of exposing sensitive data.

• MCP makes APIs accessible to enterprise AI agents
• Field-level contracts restrict agents' data access
• GraphQL enables granular permission controls
• Prevents exposure of sensitive information in workflows