---
title: "MCP maintainers chart protocol's next phase with agent auth work"
slug: "mcp-maintainers-chart-protocols-next-phase-with-agent-auth-work"
published: "2026-09-22"
beat: "Tools"
tags: ["Tools", "Business"]
creator: "Agentry Newsroom"
editor: "Susanne Sperling, Editor — Human in the Loop"
tools: ["Claude (Anthropic)", "Perplexity Sonar"]
creativeWorkStatus: "verified"
dateReviewed: "2026-09-22"
aiActArticle50: "compliant"
humanView: "https://agentry.news/tools/mcp-maintainers-chart-protocols-next-phase-with-agent-auth-work"
agentView: "https://agentry.news/agent/mcp-maintainers-chart-protocols-next-phase-with-agent-auth-work"
---# MCP maintainers chart protocol's next phase with agent auth work

> David Soria Parra and Den Delimarsky, lead maintainers of the Model Context Protocol, published a roadmap on August 22, 2026, outlining five priority areas for the standard's evolution, including prog

*Drafted by an AI agent. Verified by Susanne Sperling, Editor — Human in the Loop. [AI policy](/ai-policy).*

**MCP maintainers David Soria Parra and Den Delimarsky published a roadmap on August 22, 2026,** charting the Model Context Protocol's priorities beyond its July stateless overhaul, focusing on agent authorization, progressive tool discovery, and enterprise identity standards [Developers Digest](https://www.developersdigest.tech/blog/mcp-roadmap-progressive-discovery-agent-auth).

The roadmap identifies **five priority areas**: agentic messaging primitives, HTTP-native transport unification and hardening, agent identity and enterprise security, improved primitives, and enhanced SDK developer experience [NXCode](https://www.nxcode.io/resources/news/mcp-roadmap-agent-identity-tasks-tool-discovery-2026). These targets address operational friction points that emerge when organizations deploy agents across multiple tools and data sources at scale.

## Progressive discovery narrows tool exposure

One centerpiece is **progressive discovery**, which allows an MCP server to expose a small entry point first, then reveal more of its tool catalog as the conversation narrows [Developers Digest](https://www.developersdigest.tech/blog/mcp-roadmap-progressive-discovery-agent-auth). This reduces initial tool namespace bloat and lets agents discover relevant capabilities incrementally—critical for systems managing hundreds of tools or permissions-gated resources.

## Agent identity replaces pasted keys

The roadmap's security strand targets **standardized agent identity** verification. "We want MCP servers to have a standardized way to identify and trust these agent identities—built on existing standards rather than relying on pasted API keys and long-lived tokens," according to the maintainers' post [YAge](https://yage.ai/share/mcp-agent-identity-caller-shift-en-20260828.html). Planned mechanisms include **DPoP (Demonstration of Proof-of-Possession)**, **Workload Identity Federation**, the **ID-JAG grant**, and **token exchange** protocols—all aimed at replacing manual credential injection with cryptographic and federated identity flows [Mer.vin](https://mer.vin/news/mcps-new-roadmap-whats-next-for-the-protocol/).

These identity changes reflect a shift in how agents operate at enterprise scale: rather than embedding long-lived secrets in agent configurations, the protocol is moving toward caller verification and credential binding tied to specific workloads or agents.

## Context: July stateless revision

The roadmap builds on the **July 28, 2026 specification revision**, which made MCP servers stateless and introduced a `server/discover` handshake—a foundational change that enabled the progressive discovery work now outlined [Mind Pattern](https://mindpattern.ai/briefings/2026-08-23). That statelessness removed server-side session overhead, making MCP more suitable for ephemeral agent deployments and cloud-native architectures.

The timing reflects growing production adoption of agentic patterns: as more enterprises connect agents to CRM platforms, data warehouses, and internal tools via MCP, protocol-level primitives for discovery, authorization, and identity have moved from nice-to-have to blocking issues.