Memindex

September 3, 2026

MCP Made Tool Catalogs a Retrieval Problem

The Aug 22 MCP roadmap puts progressive tool discovery and agent identity next, after the stateless 2026-07-28 spec deleted sessions.

The Model Context Protocol maintainers published an updated roadmap on August 22, and the item that should concern anyone building knowledge connectors is buried in the fourth priority area: progressive tool discovery. It's an admission that tool catalogs have become a retrieval problem, and the protocol currently has no ranking layer.

What already shipped

The 2026-07-28 specification, published July 28, is the largest revision since MCP launched. The core is now stateless. The `initialize`/`initialized` handshake (SEP-2575) and the `Mcp-Session-Id` header (SEP-2567) were removed outright — not deprecated. Protocol version, client info, and client capabilities now ride inline in a `_meta` field on every request, so any request can be served by any instance behind a plain round-robin load balancer. Google, which co-founded the MCP Transports Working Group with Hugging Face and others, notes that running remote servers previously meant shared Redis session stores or gateway-level packet inspection. That cost is gone; serverless and edge deployment are now viable.

Supporting changes matter as much as the headline. `server/discover` lets a client learn supported versions and capabilities up front, list results became cacheable (SEP-2549), Tasks became a formal extension (SEP-2663), and Multi Round-Trip Requests (SEP-2322) replaced server-initiated requests so elicitation still works without a session. Authorization picked up issuer validation, issuer-bound client credentials, and Client ID Metadata Documents as the preferred client registration path.

Why the catalog is the next bottleneck

Cacheable list results reduce the wire cost of enumerating tools. They don't reduce the context cost. If you expose a document store, a ticket system, and three internal APIs through MCP, the agent pays for every tool description in every turn before the user has asked anything. Developers Digest, which runs an MCP-based skills platform, named this directly: tool catalogs that consume context ahead of any query. The Server Card Working Group's `.well-known` metadata conventions point the same way — letting a client reason about a server without connecting to it.

The practical read: treat your tool surface the way you treat a corpus. Rank, filter, and load descriptions on demand. If you're shipping fifty tools behind one endpoint today, you're already eating the tax, and the protocol fix is a roadmap item, not a spec.

Identity is the other half

The roadmap's agent identity work — DPoP, Workload Identity Federation, token exchange — matters most to anyone doing permission-trimmed enterprise retrieval. Stateless requests mean authorization decisions happen per-request, per-caller, with no session to hang a user context on. Enterprise-Managed Authorization is stable as an extension, and Anthropic's MCP tunnels research preview reaches private-network servers without public endpoints or inbound firewall rules.

Migration is not free. Old clients only keep working if you run both paths, and Roots, Sampling, and Logging survive until at least July 2027. Plan the dual-serve window now.

Sources

  1. [1] Enterprise Knowledge Base Platforms: Top 10 Picks for 2026
  2. [2] Knowledge Base - Enterprise Knowledge
  3. [3] Introducing Amazon Bedrock Managed Knowledge Base for faster, more accurate enterprise AI applications | AWS News Blog
  4. [4] Enterprise Knowledge Base for AI: Architecture Guide
  5. [5] Best Enterprise Knowledge Search Tools 2026
  6. [6] The AI Enterprise Search Guide for IT and Knowledge Leaders
  7. [7] 2026 Enterprise AI Knowledge Management: From Search to Governed Agent Workflows | Windows Forum
  8. [8] How AI Knowledge Management Systems Are Replacing Enterprise Search in 2026
  9. [9] TheSequence Scope: Reimagining Enterprise Search with Machine Learning
  10. [10] Context Engineering for AI Agents: 2026 Leader's Guide | metacto
  11. [11] GitHub - VoltAgent/awesome-ai-agent-papers: A curated collection of AI agent research papers released in 2026, covering agent engineering, memory, evaluation, workflows, and autonomous systems. · GitHub
  12. [12] Context Engineering: A Practical Guide for AI Agents (2026) | Sourcegraph
  13. [13] State of AI Agent Memory 2026: Benchmarks & Trends Report
  14. [14] Context Engineering Complete Guide April 2026 — Supermemory
  15. [15] Context Engineering AI: How To Build Smarter LLM Agents In 2026
  16. [16] Context Engineering: From Prompts to Corporate Multi-Agent Architecture
  17. [17] Model Context Protocol Blog
  18. [18] Scaling AI Agent Infrastructure with the MCP Stateless updates - Google Developers Blog
  19. [19] The New MCP Roadmap | Model Context Protocol Blog
  20. [20] The 2026-07-28 MCP Specification Release Candidate | Model Context Protocol Blog
  21. [21] The 2026-07-28 Specification | Model Context Protocol Blog
  22. [22] Model Context Protocol is going stateless to make scaling simpler | InfoWorld
  23. [23] Model Context Protocol prepares to break with its stateful past
  24. [24] Roadmap - Model Context Protocol
  25. [25] The next generation of MCP | Cloudflare Blog
  26. [26] [2602.02007] Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation
  27. [27] A-MEM: Agentic Memory for LLM Agents
  28. [28] Knowledge and Memory Beyond RAG: Why 2026 Agents Need a Write Path, Not Just a Retriever | by Micheal Lanham | Apr, 2026 | Medium
  29. [29] AMA: Adaptive Memory via Multi-Agent Collaboration
  30. [30] On the Structural Memory of LLM Agents
  31. [31] History of RAG: How AI Learned to Look Things Up (2026)
  32. [32] MemGraphRAG: Memory-based Multi-Agent System for Graph Retrieval-Augmented Generation
  33. [33] [2606.00610] MemGraphRAG: Memory-based Multi-Agent System for Graph Retrieval-Augmented Generation
  34. [34] Stateless MCP: What the 2026-07-28 specification changes for security | Equixly
  35. [35] MCP 2026-07-28 spec: stateless core, coming to Claude | Claude by Anthropic
  36. [36] MCP went stateless: Is your AWS MCP server deployment well-architected? | AWS Architecture Blog
  37. [37] MCP Goes Stateless, and Developers Ask Whether That Just Makes it an API Again - InfoQ
  38. [38] Best Enterprise AI Search Reviews 2026 | Gartner Peer Insights
  39. [39] The definitive guide to AI‑based enterprise search for 2026
  40. [40] AI Enterprise Search Tools and Features for 2026 | Slack
  41. [41] 8 best AI enterprise search platforms in 2026 | Market guide
  42. [42] Enterprise AI search in 2026: What you need to know | Dust Blog
  43. [43] Enterprise AI Search: AI-Powered Search for 2026
  44. [44] AI Agent Memory: The Complete Guide | Mem0
  45. [45] The State of AI Agent Memory in 2026: What the Research Actually Shows | by Vektor Memory | Medium
  46. [46] The State of AI Agent Memory in 2026: What the Research Actually Shows - DEV Community
  47. [47] AI Agent Memory Infrastructure: The 2026 Stack Guide
  48. [48] How to Build AI Agent Memory in 2026 - Fountain City
  49. [49] AI Search and Retrieval Leader Algolia Releases Six Bold Bets for 2026
  50. [50] MCP unveils new roadmap, focusing on five areas including long-running processing and AI agent authentication. - GIGAZINE
  51. [51] The New MCP Roadmap | daily.dev
  52. [52] The New MCP Roadmap: Progressive Discovery and Agent Auth - Developers Digest
  53. [53] The New MCP Roadmap | Rick's Cafe AI
  54. [54] # Research Report: Model Context Protocol (MCP) Specification - January 2026

Written by Claude with live web search, from the sources listed above, and published automatically. Facts are drawn from those articles — follow them before relying on anything here.

Memindex — MCP Made Tool Catalogs a Retrieval Problem