{
  "schema_version": "rangoon.article.v1",
  "id": "https://rangoon.ai/insights/archive-google-a2a-protocol-launch-boundaries/",
  "url": "https://rangoon.ai/insights/archive-google-a2a-protocol-launch-boundaries/",
  "markdown_url": "https://rangoon.ai/insights/archive-google-a2a-protocol-launch-boundaries/index.md",
  "json_url": "https://rangoon.ai/insights/archive-google-a2a-protocol-launch-boundaries/index.json",
  "title": "Google’s A2A launch made agent interoperability a separate boundary",
  "summary": "Google announced A2A in April 2025 to let independent agents discover capabilities and coordinate tasks, while leaving authorization and tool control to each implementation.",
  "language": "en",
  "author": {
    "name": "Rangoon Editorial",
    "url": "https://rangoon.ai/insights/editorial/"
  },
  "publisher": {
    "name": "Rangoon",
    "maintainer": "Hypler",
    "url": "https://rangoon.ai/"
  },
  "format": "News analysis",
  "topic": {
    "slug": "agents",
    "label": "Agent systems",
    "url": "https://rangoon.ai/insights/topics/agents/"
  },
  "tags": [
    {
      "slug": "google",
      "label": "Google",
      "url": "https://rangoon.ai/insights/tags/google/"
    },
    {
      "slug": "a2a",
      "label": "A2A",
      "url": "https://rangoon.ai/insights/tags/a2a/"
    },
    {
      "slug": "agent-interoperability",
      "label": "agent interoperability",
      "url": "https://rangoon.ai/insights/tags/agent-interoperability/"
    },
    {
      "slug": "protocols",
      "label": "protocols",
      "url": "https://rangoon.ai/insights/tags/protocols/"
    }
  ],
  "dates": {
    "published": "2026-10-23",
    "updated": "2026-10-23",
    "event": "2025-04-09"
  },
  "event_source_id": 1,
  "retrospective": true,
  "reading_minutes": 3,
  "takeaways": [
    "A2A’s launch positioned agents as collaborating services, with discovery and task state as shared concerns.",
    "A protocol connection can transmit a request without granting authority to perform the underlying action.",
    "Compatibility should be tested through declared capabilities, denied cases, and receipt evidence rather than a logo or protocol label."
  ],
  "source_note": "Historical reporting: Google announced A2A on April 9, 2025. The linked A2A v0.2.0 specification was released later, on June 9, 2025, and is used only to identify documented protocol concepts; it is not evidence that every later field existed at launch.",
  "sources": [
    {
      "id": 1,
      "title": "Google Developers Blog: Announcing the Agent2Agent Protocol (A2A)",
      "url": "https://developers.googleblog.com/en/a2a-a-new-era-of-agent-interoperability/",
      "published": "2025-04-09"
    },
    {
      "id": 2,
      "title": "A2A Protocol v0.2.0: Agent2Agent Protocol Specification",
      "url": "https://github.com/a2aproject/A2A/blob/v0.2.0/docs/specification.md",
      "published": "2025-06-09"
    }
  ],
  "image": {
    "url": "https://rangoon.ai/assets/editorial/archive-google-a2a-protocol-launch-boundaries-diagram-v1.webp",
    "thumbnail_url": "https://rangoon.ai/assets/editorial/archive-google-a2a-protocol-launch-boundaries-diagram-v1-thumb.webp",
    "alt": "Original explanatory diagram: A2A: remote task lifecycle. Discovery, Agent card, Task request, Status update, Artifact. Conceptual relationships, not a product interface or benchmark.",
    "type": "original-diagram",
    "caption": "A2A: remote task lifecycle. Original Rangoon explanatory diagram; conceptual relationships, not an event photograph, vendor interface, or measured result.",
    "credit": "Rangoon / Hypler",
    "source_url": "https://rangoon.ai/brand/",
    "rights_url": "https://rangoon.ai/terms/#media-use",
    "rights_note": "Hypler artwork permissions do not grant rights to third-party marks or imply endorsement."
  },
  "sections": [
    {
      "id": "launch",
      "url": "https://rangoon.ai/insights/archive-google-a2a-protocol-launch-boundaries/#launch",
      "heading": "Google introduced an agent-to-agent layer",
      "paragraphs": [
        "Google announced the Agent2Agent Protocol, or A2A, on April 9, 2025. The launch post described an open protocol for agents built by different vendors or frameworks to communicate, exchange information and coordinate actions. It named HTTP, Server-Sent Events and JSON-RPC among the standards on which the design was built, and positioned A2A as complementary to the Model Context Protocol rather than a replacement for it.",
        "The post described a client agent that formulates a task and a remote agent that attempts to perform it. It also introduced capability discovery through an Agent Card, task management with a lifecycle, messages, artifacts and user-experience negotiation. Those concepts are useful for tracing an inter-agent request. They do not make either agent trustworthy, or establish permission to use a connected system."
      ],
      "bullets": [],
      "source_ids": [
        1
      ]
    },
    {
      "id": "historical-spec",
      "url": "https://rangoon.ai/insights/archive-google-a2a-protocol-launch-boundaries/#historical-spec",
      "heading": "The early specification made the boundary concrete",
      "paragraphs": [
        "The official A2A repository’s v0.2.0 specification, released June 9, 2025, documented a client, a remote agent, an Agent Card, messages, tasks, parts and artifacts. It described HTTPS and JSON-RPC transport, streaming updates, and a server responsibility to authenticate and authorize requests under its own policies. This later version is historical technical context, not a claim about the exact April draft.",
        "Its separation is the important design clue. A capability card can describe an endpoint and advertised skills; authentication identifies a caller; authorization remains a decision by the remote service. A protocol can standardize the conversation around a task while leaving business rules, target-system entitlements and human approvals outside the wire format."
      ],
      "bullets": [],
      "source_ids": [
        1,
        2
      ]
    },
    {
      "id": "analysis-contract",
      "url": "https://rangoon.ai/insights/archive-google-a2a-protocol-launch-boundaries/#analysis-contract",
      "heading": "Analysis: make a task contract smaller than a business promise",
      "paragraphs": [
        "When two agents coordinate, write down the task contract before connecting them. Include the caller, requested outcome, allowed input fields, permitted data classification, deadline, expected artifact, idempotency key and terminal states. Do not infer that a broad capability label such as “book travel” or “update customer” authorizes every operation that could fit beneath it. The remote service should reduce the request to an exact action against an exact object.",
        "A useful test matrix includes discovery, an allowed request, an expired credential, an unauthorized target, a malformed artifact and a repeated delivery. Each case should produce an observable result and a receipt. This lets an operator distinguish a protocol failure from a business-rule denial, an external-system timeout, or a completed action whose acknowledgement was lost."
      ],
      "bullets": [],
      "source_ids": [
        1,
        2
      ]
    },
    {
      "id": "analysis-evidence",
      "url": "https://rangoon.ai/insights/archive-google-a2a-protocol-launch-boundaries/#analysis-evidence",
      "heading": "Analysis: interoperability needs auditability, not only reachability",
      "paragraphs": [
        "A successful exchange proves that two endpoints reached one another under a particular configuration. It does not prove that the downstream action was correct. Preserve a compact lineage record: the Agent Card revision used for discovery, authenticated caller, task identifier, consent or approval reference when required, target-system response and final task state. Keep request content proportionate to the task so that a debugging record does not become an unnecessary data store.",
        "The April announcement was an early interoperability milestone, not a guarantee of universal compatibility. Version changes, optional features and local authorization rules create real variation. Teams should test a named version and endpoint pair, document unsupported paths, and keep a manual recovery path when a task remains incomplete or its external outcome cannot be reconciled."
      ],
      "bullets": [],
      "source_ids": [
        1,
        2
      ]
    }
  ],
  "project_context": null,
  "related_links": [],
  "permissions_url": "https://rangoon.ai/permissions.txt",
  "content_hash": "sha256:3a7676d3fd0fbeacef2f0736ce0886223e6a709b5d2a80c26519832ee46675f2"
}