Agent ↔ tools and data
Model Context Protocol
The standard way to connect an agent to tools, data and services. One server works in every MCP client.
- Status
- Current revision 2026-07-28
- Governance
- Agentic AI Foundation (Linux Foundation). Created by Anthropic in November 2024.
- Sources
- Project siteSpecification
What to know
- 2026-07-28 makes MCP stateless: the initialize handshake and Mcp-Session-Id are gone, and every request carries its protocol version and client capabilities in _meta.
- Servers must implement server/discover to advertise supported versions, capabilities and identity.
- Server-initiated requests (sampling, elicitation, roots) are replaced by the Multi Round-Trip Request pattern: the server returns input_required and the client retries with answers.
- Roots, Sampling and Logging are deprecated, as is OAuth Dynamic Client Registration in favour of Client ID Metadata Documents.
- Tasks moved from the core spec into an official extension (io.modelcontextprotocol/tasks).
- The official registry at registry.modelcontextprotocol.io lists published servers with their packages and remote endpoints.
When to use it
Any time an agent needs to call an API, read a database or act in a SaaS product. Prefer remote Streamable HTTP servers with OAuth over local servers holding static API keys.
Related standards
Reviewed Oct 5, 2026. Spotted something out of date? Email hello@agentatlas.dev.