NEW RUNTIME

Topic hub

MCP

A New Runtime topic hub collecting signals, patterns, field notes, and public sources about mcp.

Retrieval answer

A New Runtime topic hub collecting signals, patterns, field notes, and public sources about mcp. MCP is tracked here as an evidence-linked topic, not as a static glossary entry. The page connects raw observations to pattern hypotheses, longer analysis, and public sources. Use it as the canonical landing page before drilling into individual records.

State of the topic, week of Oct 19, 2026

MCP is spreading across deployment, assistant, and app surfaces, while evidence of impact remains limited.

MCP now appears in hybrid API infrastructure, Gemini desktop, and expanded ChatGPT plugin surfaces, alongside tools for agent workflows. The period broadens where MCP can run and be encountered, but listings and documentation still do not establish adoption or impact.

What still holds

  • MCP remains a way to expose existing capabilities to agents, with deployment location changing rather than the underlying API-to-tool model. 123
  • New surfaces and integrations show possible routes to users, not proof of distribution or adoption. 456

New this fortnight

  • Apigee hybrid added native MCP support, placing the agent-facing adapter in a customerโ€™s own cluster. 12
  • Gemini Spark is beginning to roll out custom MCP support, offering user-selected integrations beyond named app connections. 76
  • OpenAI documentation describes MCP extensions across ChatGPTโ€™s sidebar, conversation, settings, file, and desktop composer surfaces, alongside a ZIP-based review workflow. 84910
  • OpenClaw moved installed-skill discovery to a permission-aware search-and-read path and supports MCP Apps alongside conversations. 1112

Weakening

  • The idea that MCP integrations are confined to conversation-oriented or centrally named app surfaces is weakened by custom integrations and broader plugin placements. 74610

Open questions

  • Will these newly described surfaces and deployment options see meaningful adoption?
  • Will the documented plugin review workflow change review speed or acceptance outcomes?
  • How broadly will custom MCP support roll out in Gemini Spark?

Watch next: Watch for evidence of rollout breadth, actual use, and operational outcomes for MCP integrations across these surfaces.

  1. Announcing MCP support in Apigee hybrid: agentic tools that run in your own cluster signal, Oct 20, 2026
  2. Announcing MCP support in Apigee hybrid: agentic tools that run in your own cluster post, Oct 20, 2026
  3. Announcing MCP support in Apigee hybrid: agentic tools that run in your own cluster - Google Developers Blog archive
  4. Build richer, more interactive plugins for ChatGPT. signal, Oct 8, 2026
  5. A plugin surface is not a distribution channel analysis, Oct 8, 2026
  6. Google seems to be rolling out support for custom MCPs on Gemini desktop app to more users. post, Oct 16, 2026
  7. Google seems to be rolling out support for custom MCPs on Gemini desktop app to more users. signal, Oct 16, 2026
  8. OpenAIโ€™s plugin submission documentation describes a workflow in which developers upload a ZIP signal, Oct 8, 2026
  9. OpenAIโ€™s plugin submission documentation describes a workflow in which developers upload a ZIP post, Oct 8, 2026
  10. Build richer, more interactive plugins for ChatGPT. post, Oct 8, 2026
  11. OpenClaw released v2026.10.1 with installed-skill discovery, MCP Apps alongside conversations signal, Oct 12, 2026
  12. OpenClaw released v2026.10.1 with installed-skill discovery, MCP Apps alongside conversations post, Oct 12, 2026

Written automatically by the New Runtime topic desk from the fortnight's published items and the archive; rewritten weekly.

Last 14 days

Latest on this topic

72 items on MCP published in the last 14 days, newest first. Showing the newest 8.

  1. Letta Code v0.34.26 adds Cursor coding workers and fixes workflow worker-run lookupField note

  2. Pydantic AI v2.62.0 adds mid-conversation thinking-effort changes, new model error types and prefixed hosted-MCP tool namesField note

  3. Staygreen uptime monitor for agent-built apps goes open source with an MCP serverField note

  4. Upstash ships MCP Tasks support in its MCP toolkit, backed by Redis and QStash, and documents Claude Code limitsField note

  5. NVIDIA released Auto Ontology , an Apache-2.0 framework for building semantic graph layers over enterprise databases.Field note

  6. Parallel Decisions API: web-grounded classificationField note

  7. Step 5 Preview plugs into Claude Code, but nothing published yet shows it is a drop-in swapAnalysis

  8. OpenAI's approval rule for gateways makes the model router the riskiest part of your stack, so keep your ability to switch models in MCP and skillsAnalysis

Pattern memory

What patterns are emerging?

1 patterns
  1. medium

    Agent protocols become the interoperability layer

    The durable integration surface for agentic software is becoming a small family of protocols that separate models and harnesses from tools, services, interfaces, and other agents.

Field notes

What should readers understand next?

11 notes

  1. Cline Hooks Put Deterministic Rules Inside The Agent Loop

    Cline's plugin hooks show how an agent harness can journal every run and block dangerous tool calls without waiting for the model to choose a guardrail.

  2. CopilotKit Brings MCP Tool Calls Into The React Interface

    CopilotKit's React integration registers MCP servers with the chat runtime, exposes their tools to the model, and renders tool-call state inside the product interface.

  3. Anthropic Moves Large Tool Libraries Out Of Context

    Anthropic's Tool Search Tool, Programmatic Tool Calling, and Tool Use Examples separate discovery, orchestration, and usage guidance for agents with large tool libraries.

  4. Dr. Skill Audits What An Agent Loads Before It Works

    Dr. Skill scans skills and MCP servers for collisions, duplication, secrets, drift, missing metadata, and unused loadout, with local and CI-friendly commands.

  5. MCP 2026 Makes The Protocol Stateless

    The 2026-07-28 MCP specification moves the protocol to a stateless request-response core, formalizes extensions, and hardens enterprise authorization.

  6. Rogue Turns Agent Risk Into A Test Harness

    Qualifire's Rogue repository exposes agent hardening as automatic evaluation and red teaming across protocols such as A2A, MCP, and Python entrypoints.

  7. Stateless MCP Makes Small, Auditable Agent Tools Practical Again

    Simon Willison's mcp-explorer, datasette-mcp, and llm-mcp-client show how the stateless specification lowers implementation cost and narrows agent capabilities.

  8. Firecrawl MCP Turns Web Search into a Bounded Agent Capability

    Firecrawl's MCP launch points to a cleaner web-context surface for agents: OAuth for humans, API-key headers for server jobs, and keyless trials for low-friction testing.

  9. Mem0 Shows Agent Memory Should Not All Live in Context

    Mem0's Claude Code experiment separates durable memory from the conversation window: retrieve the relevant slice, survive /clear, and avoid loading every memory file up front.

  10. Vercel's mcp-handler 2.0 Makes MCP Servers More Stateless

    mcp-handler 2.0 supports the 2026-07-28 MCP spec while keeping older Streamable HTTP clients on the same endpoint and removing Redis or session-storage requirements.

  11. Unabyss Builds a Shared Context Layer Across Claude, Codex, and Cursor

    Unabyss points to where agent tooling is going: context stops being one client's internal memory and becomes a separate product layer with governance and portability.

Raw signals

What changed recently?

82 signals

  1. Firecrawl MCP Cuts Context Use for Web Tools

    Firecrawl's MCP docs expose web search, scrape, parse, and interact work through bounded agent access paths for OAuth users, servers, and keyless trials.

  2. Vercel Developers: ๐š–๐šŒ๐š™-๐š‘๐šŠ๐š—๐š๐š•๐šŽ๐š› 2.0.0 is now available.

    A public X post from Vercel Developers with a linked primary source flags ๐š–๐šŒ๐š™-๐š‘๐šŠ๐š—๐š๐š•๐šŽ๐š› 2.0.0 is now available. โ–ช๏ธŽ Native support for the latest MCP spec โ–ช๏ธŽ Stateless fallback for 2025-era clients โ–ช๏ธŽ Zero session storage requirements

  3. Cursor: Today we're launching Cursor Start, a new โ‚น649/month plan for developers in India.

    A public X post from Cursor with a linked primary source flags Today we're launching Cursor Start, a new โ‚น649/month plan for developers in India. Start includes generous access to Grok 4.5 and Composer, so you can plan, build, test, and ship...

  4. langfuse.com: manage dashboards via API, CLI, MCP & Langfuse Assistant create and manage dashboards and...

    A public X post from langfuse.com with a linked primary source flags manage dashboards via API, CLI, MCP & Langfuse Assistant create and manage dashboards and widgets via the public API, let agents build them through the MCP server or CLI, or ask t...

  5. A local agent is a runtime, not one model

    A local agent stack combines models, orchestration, memory, skills, MCP tools, permissions, judges, and output guards.

  6. Uber built a platform layer for thousands of agents

    Uber's internal approach focuses on common protocols, evaluation, identity, observability, and policy instead of one mandated agent framework.

  7. Agent-ready SaaS exposes primitives to external agents

    Cloudflare's Code Mode compresses broad API access into search and execute primitives instead of thousands of tool definitions.

  8. Agentic Resource Discovery Indexes Skills, MCP, and Agents

    Hugging Face treats agent capabilities as searchable resources, creating a discovery layer for tools that can be loaded at runtime.

  9. GitHub / zapier/AutomationBench: Verification Bandwidth

    The archive captures GitHub / zapier/AutomationBench as a dated public record from GitHub / zapier/AutomationBench. It documents evaluation, review, and observability becoming the bottleneck after generation accelerates and is retained as pressure-testing evidence for the verification bandwidth trend.

  10. OpenAI / Workspace Agents: Ai-Native Operating Models

    The archive captures OpenAI / Workspace Agents as a dated public record from OpenAI / Workspace Agents. It documents AI adoption shifting jobs, coordination, review, and organizational capacity and is retained as supporting evidence for the AI-native operating models trend.

  11. Venturebeat / Salesforce Launches Headless To Turn Its Entire: Agent Protocol Interoperability

    The archive captures Venturebeat / Salesforce Launches Headless To Turn Its Entire as a dated public record from Venturebeat / Salesforce Launches Headless To Turn Its Entire. It documents agent tools and services converging on explicit interoperability protocols and is retained as branch-opening evidence for the agent protocol interoperability trend.

  12. OpenAI / Next Evolution Of Agents Sdk: Agent-Ready Software

    The archive captures OpenAI / Next Evolution Of Agents Sdk as a dated public record from OpenAI / Next Evolution Of Agents Sdk. It documents software exposing explicit capabilities, permissions, and machine-readable actions and is retained as pressure-testing evidence for the agent-ready software trend.

  13. GitHub / langchain-ai/deepagents: Agent-Ready Software

    The archive captures GitHub / langchain-ai/deepagents as a dated public record from GitHub / langchain-ai/deepagents. It documents software exposing explicit capabilities, permissions, and machine-readable actions and is retained as pressure-testing evidence for the agent-ready software trend.

  14. Agentation: Task-Specific Ai Interfaces

    The archive captures Agentation as a dated public record from Agentation. It documents model output moving from plain answers into generated task interfaces and actions and is retained as supporting evidence for the task-specific AI interfaces trend.

  15. GitHub / jgraph/drawio-mcp: Executable Design Context

    The archive captures GitHub / jgraph/drawio-mcp as a dated public record from GitHub / jgraph/drawio-mcp. It documents design systems becoming machine-readable context, constraints, and review loops and is retained as supporting evidence for the executable design context trend.

  16. Chrlschn / Mcp Is Dead Long Live Mcp: Agent Protocol Interoperability

    The archive captures Chrlschn / Mcp Is Dead Long Live Mcp as a dated public record from Chrlschn / Mcp Is Dead Long Live Mcp. It documents agent tools and services converging on explicit interoperability protocols and is retained as branch-opening evidence for the agent protocol interoperability trend.

  17. Cursor: Agent-Ready Software

    The archive captures Cursor as a dated public record from Cursor. It documents software exposing explicit capabilities, permissions, and machine-readable actions and is retained as pressure-testing evidence for the agent-ready software trend.

  18. GitHub / Agent-Field/agentfield: Agent-Ready Software

    The archive captures GitHub / Agent-Field/agentfield as a dated public record from GitHub / Agent-Field/agentfield. It documents software exposing explicit capabilities, permissions, and machine-readable actions and is retained as pressure-testing evidence for the agent-ready software trend.

  19. GitHub / msitarzewski/agency-agents: Verification Bandwidth

    The archive captures GitHub / msitarzewski/agency-agents as a dated public record from GitHub / msitarzewski/agency-agents. It documents evaluation, review, and observability becoming the bottleneck after generation accelerates and is retained as supporting evidence for the verification bandwidth trend.

  20. Google AI / Gemini Api: Agent Protocol Interoperability

    The archive captures Google AI / Gemini Api as a dated public record from Google AI / Gemini Api. It documents agent tools and services converging on explicit interoperability protocols and is retained as branch-opening evidence for the agent protocol interoperability trend.

  21. GitHub / jayminwest/overstory: Verification Bandwidth

    The archive captures GitHub / jayminwest/overstory as a dated public record from GitHub / jayminwest/overstory. It documents evaluation, review, and observability becoming the bottleneck after generation accelerates and is retained as pressure-testing evidence for the verification bandwidth trend.

  22. MCP: Model Context Protocol - Official Documentation

    The archive captures MCP: Model Context Protocol - Official Documentation as a dated public record from GitHub / a2aproject/A2A. It documents agent tools and services converging on explicit interoperability protocols and is retained as branch-opening evidence for the agent protocol interoperability trend.

  23. Cursor: Ai-Native Operating Models

    The archive captures Cursor as a dated public record from Cursor. It documents AI adoption shifting jobs, coordination, review, and organizational capacity and is retained as pressure-testing evidence for the AI-native operating models trend.

  24. Anthropic / Claude Sonnet: Routable Model Components

    The archive captures Anthropic / Claude Sonnet as a dated public record from Anthropic / Claude Sonnet. It documents models becoming replaceable or specialized components inside a more durable runtime and is retained as supporting evidence for the routable model components trend.

Next reads

Read next

Open the graph
  1. 01related materialAgent protocols become the interoperability layerContinue through the MCP topic.
  2. 02related materialCline Hooks Put Deterministic Rules Inside The Agent LoopContinue through the MCP topic.
  3. 03related materialCopilotKit Brings MCP Tool Calls Into The React InterfaceContinue through the MCP topic.
  4. 04related materialAnthropic Moves Large Tool Libraries Out Of ContextContinue through the MCP topic.
  5. 05related materialDr. Skill Audits What An Agent Loads Before It WorksContinue through the MCP topic.
Who read this page?Machine requests, hidden until opened

Loading the privacy-safe route aggregateโ€ฆ

Open the JSON contract