Signal SentrySearchUpdated Nov 20, 05:30 UTCAM Drop
Frontier watch · Protocols & interop

MCP and A2A: the agent plumbing

Are MCP and A2A settling into the stable plumbing between agents and tools, and who runs them in production?

Why it matters

If agent-to-tool (MCP) and agent-to-agent (A2A) links become neutral, versioned standards, builders can wire agents across vendors once instead of writing custom glue for every pairing, and buyers can switch platforms with less lock-in.

State of play

Newest first. Each claim is our summary of the linked primary source, with a short quote from it.

  1. ShippingOct 15, 2026Source

    Microsoft says Agent-to-Agent (A2A) support in Microsoft Foundry is generally available, in its September 2026 Foundry update.

    Tool Search and Agent-to-Agent (A2A) are GA.

    What it does not show: A monthly roundup; it does not detail which A2A versions or features the GA covers or name customers.

  2. DocumentedOct 9, 2026Source

    Sierra published a draft of Personal Agent Protocol ("Poppy"), developed with Meta: a well-known discovery file, OAuth sign-in and one session token that lets a personal agent work with a company within access the customer approved.

    Poppy is built on the principle that customers decide what access to give their personal agents, and companies set parameters for what those agents can do.

    What it does not show: A draft for comment; the reference implementation is still to come, and the 35 added design partners join the design process, not production use. Payments are left to extensions.

  3. AnnouncedAug 27, 2026Source

    The A2A project says A2A was accepted as a Growth Stage project at the Agentic AI Foundation, the Linux Foundation fund that already hosts MCP.

    The Agent2Agent (A2A) protocol has officially been accepted as a Growth Stage project at the Agentic AI Foundation (AAIF).

    What it does not show: Its adoption figures (over 150 organizations, production use) are the project's own and name no specific production deployments.

  4. DocumentedJul 28, 2026Source

    AWS says its AgentCore Gateway can advertise MCP 2026-07-28 alongside 2025-11-25, so clients on the older version are unaffected while new clients opt in.

    Adding 2026-07-28 to a gateway that also advertises 2025-11-25 does not change anything for clients that request the older version.

    What it does not show: Vendor post about its own gateway; no named customer running the new version is cited.

  5. DocumentedJul 28, 2026Source

    MCP's maintainers published spec version 2026-07-28, which turns MCP into a stateless request/response protocol by removing the handshake and sessions, a breaking change.

    The highlight of this release is a stateless protocol core - MCP is transforming from a bidirectional stateful protocol into a request/response stateless protocol.

    What it does not show: Spec and Tier 1 SDK release; it does not show how many deployed clients and servers have moved to the new version.

  6. DocumentedMar 12, 2026Source

    The A2A project released v1.0 of its agent-to-agent spec, describing it as a maturity release that clarifies ambiguous areas and addresses enterprise deployment needs.

    The v1.0 release emphasizes maturity rather than reinvention: the core ideas remain intact, while rough edges have been removed, ambiguous areas clarified, and enterprise deployment requirements addressed more directly.

    What it does not show: The project's own post; v1.0 includes breaking changes to the interaction protocol, per the same post.

  7. AnnouncedDec 9, 2025Source

    MCP's maintainers announced that Anthropic is donating MCP to the new Agentic AI Foundation, a directed fund under the Linux Foundation, as a founding project.

    Anthropic is donating MCP to the Agentic AI Foundation, a directed fund under the Linux Foundation.

    What it does not show: A governance move; the post says maintainers keep technical direction, so it does not by itself change the protocol.

What would change the picture

Signals we are watching for. Our own reading of where the thread could go: not predictions, and not sourced claims.

  • A second MCP spec release ships without breaking changes under the new feature lifecycle and deprecation policy Would suggest: MCP's core is stabilizing and upgrades become routine rather than migrations
  • Another major cloud or SaaS vendor declares A2A v1.0 generally available and names production customers using it Would suggest: agent-to-agent delegation moving from pilots into production infrastructure
  • The Tasks extension graduates from an MCP extension into the core specification Would suggest: long-running agent work is settling on MCP rather than a separate protocol
  • The IETF agentproto working group adopts a dialog management draft that MCP or A2A says it will carry Would suggest: agent protocols converging with IETF standards rather than staying foundation-only

From the Drops

SecurityResearch

Preprint: A2A Agent Cards can be rigged to sway which agent an orchestrator picks

Researchers probed 36.1 million domains, found 310 deployed Agent Cards and built SECCARD. Its best LLM judge caught 97.0% of injected attacks at a 1.7% false-positive rate; on another test nine judges ranged 25.8% to 92.3%.

Why it matters: Text in an agent's public description can steer selection before any call is made. Treat Agent Card fields as untrusted input, and vet the judge model you rely on.

Preliminary Way in: attacker-written text in an A2A Agent Card that sways which agent gets chosen

Controls that would have helped:

From Drop #092 · Tue, Nov 17, 2026 · Evening
AgentsShipping

Duco cut agent context from 200,000 to about 20,000 tokens by loading tools on demand

AWS says Duco, which serves 200+ financial clients, exposed 300–400 tools per tenant, filling context before work began. Three meta-tools now find, load and call tools only when needed; AWS reports better answer quality.

Why it matters: Large MCP tool catalogs can quietly degrade agents. Measure tool-description tokens per session and, as Duco did, run evals before deciding an agent has gotten worse.

From Drop #091 · Tue, Nov 17, 2026 · Morning
AgentsDocumented

AWS shows A2A agents working across two companies' accounts with no shared secrets

AWS's walkthrough links a LangChain orchestrator and a Strands SaaS agent in separate accounts over A2A, with OAuth both ways and only public values swapped at onboarding. The SaaS agent publishes one curated skill.

Why it matters: Vendors exposing agents to customers get a documented path using each side's own identity provider. Check what your Agent Card reveals about internal tools.

From Drop #082 · Thu, Nov 12, 2026 · Evening
AgentsAnnounced

MCP moves to two spec releases a year; next version targeted for March 30, 2027

The MCP maintainers say spec releases will generally land on the last Tuesday of March and September, with soft and hard freezes and SDK betas before each. Proposals that are not ready move to a later release.

Why it matters: Teams building on MCP can plan upgrades around fixed dates. Mark the 2027-03-30 version and test SDK betas against your servers when they appear.

From Drop #081 · Thu, Nov 12, 2026 · Morning
AgentsAnnounced

MCP maintainers draft a file upload and download extension, still experimental

The MCP blog says a working group drafted io.modelcontextprotocol/files: files move outside JSON-RPC, the model never sees the bytes or URLs, and a credentials flag limits where the MCP token goes. No reference implementation yet.

Why it matters: If you build MCP servers, hold off on bespoke file tools. The draft warns a malicious server could still steer uploads and suggests hosts allowlist target origins.

From Drop #080 · Wed, Nov 11, 2026 · Evening
AgentsShipping

Workday says its agent platform cut prototype time from 4 to 5 months to under a day

In a LangChain case study, Workday says a team of 4 built an agent server, eval platform and SDK on LangGraph and LangSmith, with audit and data controls built in. The team is now roughly 30 people.

Why it matters: The speedup came from one shared platform that handles auth, compliance and protocols for every team. It is a vendor case study, so there are no independent figures.

From Drop #078 · Tue, Nov 10, 2026 · Evening
AgentsDocumented

Amazon team puts an MCP gateway between agents and its data lake, read-only by design

AWS says a team serving Amazon's North America Sort Center planners built one MCP server on ECS Fargate: new datasets need only a JSON metadata file, queries are SELECT-only, and every tool call is logged centrally.

Why it matters: Access rules live in one gateway instead of in each agent. If your agents query data directly, check where permissions are enforced and whether writes are blocked at the tool.

From Drop #073 · Sun, Nov 8, 2026 · Morning
AgentsShipping

U Mobile says its helpdesk agents cut trouble tickets by up to 84.5% in key categories

AWS's post says Malaysian carrier U Mobile runs 8 agents over A2A and MCP across more than 150 scenarios. About 70 percent of requests resolve by keyword matching with no model call, and staff confirm changes to live accounts.

Why it matters: A production pattern worth copying: cheap fixed routing first, models only when needed, and a person signing off on account changes. The ticket figures are U Mobile's own measurements.

From Drop #067 · Thu, Nov 5, 2026 · Morning
SecurityResearch

Preprint finds the official MCP SDKs skip some spec checks that servers inherit

Researchers report in a preprint that their MCPGuard tool confirmed 13 conformance gaps in the official Python 1.26.0 MCP SDK and 19 in TypeScript 1.29.0, none caught by two black-box baselines, and that they can carry into open servers.

Why it matters: Building on an official MCP SDK does not mean your server enforces the spec. Add your own input and session validation, and watch SDK releases for fixes.

From Drop #066 · Wed, Nov 4, 2026 · Evening
AgentsResearch

Preprint: MCP and CLI interfaces with matched capabilities change agent behavior

Across six models and three environments, researchers report that matched MCP and CLI interfaces produced substantial differences in task performance and cost, with structured divergence in 94.3% of 2,004 episodes.

Why it matters: The authors say their DART repair method raised scores by up to 9.0 points and cut MCP token cost by up to 27.1k tokens per task: how tools are presented is not neutral.

From Drop #056 · Fri, Oct 30, 2026 · Evening
AgentsResearch

Preprint: the same tools via MCP or a CLI changed agent behavior in 94.3% of episodes

Across six models and three environments, researchers report matched interfaces gave different scores and costs. Their DART repair raised scores by up to 9.0 points and cut MCP token cost by up to 27.1k tokens per task.

Why it matters: MCP versus CLI isn't a neutral choice. Test both on your own tasks before standardizing on one.

From Drop #049 · Tue, Oct 27, 2026 · Morning
AgentsResearch

Preprint compares MCP and CLI for agent tools and recommends a hybrid

Across 14 evaluation dimensions, the authors find CLI more efficient for single-user, local workflows, while MCP covers needs CLI cannot: fine-grained access control, credential isolation, structured audit and multi-tenant governance.

Why it matters: The authors argue progressive disclosure brings MCP's token use close to CLI's, so production teams need not trade governance for context budget.

From Drop #042 · Fri, Oct 23, 2026 · Evening
AgentsDocumented

Anthropic gives Agent Skills to the Agentic AI Foundation, now home to 10 projects

The Linux Foundation says AAIF added Agent Skills from Anthropic, Open Responses from OpenAI, OpenEnv from Hugging Face and OpenSandbox from Alibaba, joining MCP and A2A. It says AAIF now has 305 members.

Why it matters: Skills, a cross-provider API spec, RL environments and sandboxes now sit under neutral governance, so builders can adopt them with less single-vendor risk.

From Drop #041 · Fri, Oct 23, 2026 · Morning
AgentsShipping

Eon cut engineers' tool setup from two days to two minutes with one MCP gateway

In an AWS post co-written with Eon, the company says it swapped per-tool credentials and local MCP servers for one AgentCore gateway across AWS, GCP and Azure; its SRE agent gets the same tools with different permissions.

Why it matters: It is a working pattern for giving people and agents one governed door to tools, with Cedar policies deciding which tools each person or agent may call.

From Drop #037 · Wed, Oct 21, 2026 · Morning
AgentsResearch

GatewayBench preprint tests seven MCP gateways on 524 tasks with 11,118 real tools

Researchers built tasks across six domains from 251 MCP servers, injecting irrelevant tools to mimic large deployments. They report substantial gaps between gateway designs in task completion, efficiency and resilience to attacks.

Why it matters: The authors say gateway effects interact strongly with the model and harness, so the middleware between agents and tools deserves its own testing.

From Drop #033 · Mon, Oct 19, 2026 · Morning