[{"data":1,"prerenderedAt":331},["ShallowReactive",2],{"blog-post-blog_en-agent2agent-protokoll-fuer-unternehmen":3},{"id":4,"title":5,"body":6,"cover":315,"date":316,"description":317,"draft":318,"extension":319,"meta":320,"navigation":321,"path":322,"seo":323,"stem":324,"tags":325,"__hash__":330},"blog_en\u002Fen\u002Fblog\u002Fagent2agent-protokoll-fuer-unternehmen.md","Agent2Agent Protocol for Enterprises: Introducing A2A Safely",{"type":7,"value":8,"toc":310},"minimark",[9,13,18,21,24,53,56,60,63,66,98,287,290,294,297,306],[10,11,12],"p",{},"The Agent2Agent Protocol is moving into architecture discussions because companies no longer build only isolated AI agents. They want agents from different products, teams, and partners to coordinate work. A2A v1.0 promises an open standard for this, but it does not solve accountability, data, or operations questions by itself.",[14,15,17],"h2",{"id":16},"what-the-agent2agent-protocol-changes","What the Agent2Agent Protocol Changes",[10,19,20],{},"A2A does not standardise how an agent accesses a tool. In many architectures, that is the role of MCP. The Agent2Agent Protocol describes communication between agents: discovering capabilities, delegating tasks, tracking status, and exchanging results.",[10,22,23],{},"For growing software companies, this changes four things:",[25,26,27,35,41,47],"ul",{},[28,29,30,34],"li",{},[31,32,33],"strong",{},"Agent discovery:"," Agent Cards make visible which skills an agent offers, which interfaces it supports, and which authentication is required.",[28,36,37,40],{},[31,38,39],{},"Task model:"," A2A treats agent work as a task with status, streaming, cancellation, and result artefacts, not just a single API call.",[28,42,43,46],{},[31,44,45],{},"Interoperability:"," HTTP+JSON, JSON-RPC, and gRPC fit existing platforms better than proprietary agent bridges.",[28,48,49,52],{},[31,50,51],{},"Governance pressure:"," When agents operate across team or company boundaries, identity, authorisation, auditability, and cost control become architecture concerns.",[10,54,55],{},"The value is therefore not in making as many agents as possible talk to each other. Value appears when clearly defined agents collaborate under control without every team inventing its own integration logic.",[14,57,59],{"id":58},"where-teams-should-start-before-a2a-adoption","Where Teams Should Start Before A2A Adoption",[10,61,62],{},"The most common mistake is treating A2A like a new integration library. As soon as one agent calls another agent, a chain of trust, data sharing, and business accountability appears.",[10,64,65],{},"Before a production A2A pilot, teams should clarify at least:",[25,67,68,74,80,86,92],{},[28,69,70,73],{},[31,71,72],{},"Visibility:"," Which Agent Cards may be public, internal, or discoverable only through a curated registry?",[28,75,76,79],{},[31,77,78],{},"Identity:"," Does the calling agent act on behalf of a user, a team, or a technical service?",[28,81,82,85],{},[31,83,84],{},"Skill scope:"," Which skills are read-only, and which may trigger actions in backend systems?",[28,87,88,91],{},[31,89,90],{},"Operations:"," Which task IDs, trace IDs, costs, and error states are logged?",[28,93,94,97],{},[31,95,96],{},"Approval boundaries:"," When does an agent need human approval instead of automatically delegating to the next agent?",[99,100,105],"pre",{"className":101,"code":102,"language":103,"meta":104,"style":104},"language-json shiki shiki-themes github-light github-dark","{\n  \"name\": \"Supplier Risk Agent\",\n  \"supportedInterfaces\": [{ \"url\": \"https:\u002F\u002Fagents.example.com\u002Fa2a\u002Fv1\", \"protocolBinding\": \"HTTP+JSON\", \"protocolVersion\": \"1.0\" }],\n  \"capabilities\": { \"streaming\": true, \"extendedAgentCard\": true },\n  \"securitySchemes\": { \"oidc\": { \"openIdConnectSecurityScheme\": { \"openIdConnectUrl\": \"https:\u002F\u002Fid.example.com\u002F.well-known\u002Fopenid-configuration\" } } },\n  \"security\": [{ \"oidc\": [\"supplier-risk.read\"] }],\n  \"skills\": [{ \"id\": \"assess_supplier_risk\", \"name\": \"Assess supplier risk\" }]\n}\n","json","",[106,107,108,117,134,175,204,233,252,281],"code",{"__ignoreMap":104},[109,110,113],"span",{"class":111,"line":112},"line",1,[109,114,116],{"class":115},"sVt8B","{\n",[109,118,120,124,127,131],{"class":111,"line":119},2,[109,121,123],{"class":122},"sj4cs","  \"name\"",[109,125,126],{"class":115},": ",[109,128,130],{"class":129},"sZZnC","\"Supplier Risk Agent\"",[109,132,133],{"class":115},",\n",[109,135,137,140,143,146,148,151,154,157,159,162,164,167,169,172],{"class":111,"line":136},3,[109,138,139],{"class":122},"  \"supportedInterfaces\"",[109,141,142],{"class":115},": [{ ",[109,144,145],{"class":122},"\"url\"",[109,147,126],{"class":115},[109,149,150],{"class":129},"\"https:\u002F\u002Fagents.example.com\u002Fa2a\u002Fv1\"",[109,152,153],{"class":115},", ",[109,155,156],{"class":122},"\"protocolBinding\"",[109,158,126],{"class":115},[109,160,161],{"class":129},"\"HTTP+JSON\"",[109,163,153],{"class":115},[109,165,166],{"class":122},"\"protocolVersion\"",[109,168,126],{"class":115},[109,170,171],{"class":129},"\"1.0\"",[109,173,174],{"class":115}," }],\n",[109,176,178,181,184,187,189,192,194,197,199,201],{"class":111,"line":177},4,[109,179,180],{"class":122},"  \"capabilities\"",[109,182,183],{"class":115},": { ",[109,185,186],{"class":122},"\"streaming\"",[109,188,126],{"class":115},[109,190,191],{"class":122},"true",[109,193,153],{"class":115},[109,195,196],{"class":122},"\"extendedAgentCard\"",[109,198,126],{"class":115},[109,200,191],{"class":122},[109,202,203],{"class":115}," },\n",[109,205,207,210,212,215,217,220,222,225,227,230],{"class":111,"line":206},5,[109,208,209],{"class":122},"  \"securitySchemes\"",[109,211,183],{"class":115},[109,213,214],{"class":122},"\"oidc\"",[109,216,183],{"class":115},[109,218,219],{"class":122},"\"openIdConnectSecurityScheme\"",[109,221,183],{"class":115},[109,223,224],{"class":122},"\"openIdConnectUrl\"",[109,226,126],{"class":115},[109,228,229],{"class":129},"\"https:\u002F\u002Fid.example.com\u002F.well-known\u002Fopenid-configuration\"",[109,231,232],{"class":115}," } } },\n",[109,234,236,239,241,243,246,249],{"class":111,"line":235},6,[109,237,238],{"class":122},"  \"security\"",[109,240,142],{"class":115},[109,242,214],{"class":122},[109,244,245],{"class":115},": [",[109,247,248],{"class":129},"\"supplier-risk.read\"",[109,250,251],{"class":115},"] }],\n",[109,253,255,258,260,263,265,268,270,273,275,278],{"class":111,"line":254},7,[109,256,257],{"class":122},"  \"skills\"",[109,259,142],{"class":115},[109,261,262],{"class":122},"\"id\"",[109,264,126],{"class":115},[109,266,267],{"class":129},"\"assess_supplier_risk\"",[109,269,153],{"class":115},[109,271,272],{"class":122},"\"name\"",[109,274,126],{"class":115},[109,276,277],{"class":129},"\"Assess supplier risk\"",[109,279,280],{"class":115}," }]\n",[109,282,284],{"class":111,"line":283},8,[109,285,286],{"class":115},"}\n",[10,288,289],{},"For internal agents, a private registry with clear owners is often enough at the beginning. For partner or supply-chain processes, signed Agent Cards, API management, rate limits, mTLS, OAuth2 or OpenID Connect, and W3C Trace Context quickly become important.",[14,291,293],{"id":292},"why-this-matters","Why This Matters",[10,295,296],{},"The Agent2Agent Protocol can reduce integration cost because agents no longer have to cooperate only inside a single product or framework. At the same time, risk increases when poorly defined agents pass tasks, data, and decisions across system boundaries.",[10,298,299,300,305],{},"For founders, product leaders, and engineering managers, A2A is therefore not just an AI topic. It is an architecture decision with consequences for operating cost, compliance, vendor lock-in, and accountability. Teams that classify A2A early can test agentic workflows under control. Teams that introduce it without governance merely build a hard-to-audit automation layer on top of existing interface problems. An ",[301,302,304],"a",{"href":303},"\u002Fen\u002F#packages","Architecture & AI Review"," can assess whether agents, MCP integrations, and A2A interfaces form a viable operating model.",[307,308,309],"style",{},"html pre.shiki code .sVt8B, html code.shiki .sVt8B{--shiki-default:#24292E;--shiki-dark:#E1E4E8}html pre.shiki code .sj4cs, html code.shiki .sj4cs{--shiki-default:#005CC5;--shiki-dark:#79B8FF}html pre.shiki code .sZZnC, html code.shiki .sZZnC{--shiki-default:#032F62;--shiki-dark:#9ECBFF}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"title":104,"searchDepth":119,"depth":119,"links":311},[312,313,314],{"id":16,"depth":119,"text":17},{"id":58,"depth":119,"text":59},{"id":292,"depth":119,"text":293},null,"2026-05-22","The Agent2Agent Protocol makes AI agents interoperable. What companies should clarify before A2A adoption, governance and operations.",false,"md",{},true,"\u002Fen\u002Fblog\u002Fagent2agent-protokoll-fuer-unternehmen",{"title":5,"description":317},"en\u002Fblog\u002Fagent2agent-protokoll-fuer-unternehmen",[326,327,328,329],"AI","Software Architecture","Engineering Leadership","Governance","DLlTyWzo4E79_w6b2_XyQmPSn0eYgCwTCIZUfgj5wEU",1780122461647]