Agent Card
A public, machine-readable résumé: identity, interfaces, skills, media types, security and extensions.
From first handshake to finished task. A practical, visual guide to the open protocol that lets independent AI agents discover, coordinate and deliver work—without exposing how they think.
A2A standardizes the boundary between agents. It says how to advertise capability, exchange typed content, manage work over time and report results—not how either agent must be built.
Fetch the public Agent Card. It declares identity, skills, interfaces, media types, authentication and optional extensions.
Think of A2A as a professional hand-off between two capable teams. Each team stays autonomous; the contract makes the hand-off observable and interoperable.
A public, machine-readable résumé: identity, interfaces, skills, media types, security and extensions.
A turn in the conversation, with a user or agent role and one or more typed parts.
A stateful unit of work with an ID, context, status, history and optional artifacts.
The atomic content unit: text, URL, raw bytes or structured data. In v1, the populated member is the discriminator.
A durable result of the work: a report, dataset, map, file or any ordered collection of parts.
A concrete URL plus binding and version. JSON-RPC, gRPC and HTTP+JSON are standard bindings.
Who can do this work?
Which tools and context can I use?
Use both: an orchestrator can delegate a specialist task over A2A; that specialist can use MCP internally to reach databases, files and tools.
A short answer may return as a Message. Anything that needs identity, progress, resumption or delivery over time becomes a Task.
The task is active. Observe it through SSE, polling or a push notification configuration.
SendMessage for work that can complete inside one HTTP exchange.
SendStreamingMessage or SubscribeToTask over server-sent events.
Register a secured webhook for disconnected or long-running clients.
SendMessageSendStreamingMessageGetTaskListTasksCancelTaskSubscribeToTaskCreateTaskPushNotificationConfigGetTaskPushNotificationConfigListTaskPushNotificationConfigsDeleteTaskPushNotificationConfigGetExtendedAgentCardVersion 1.0 made interoperability stricter and security more explicit. The 1.0.1 maintenance release tightened media types, task-state definitions and error transcoding.
Google and more than 50 launch partners publish A2A as an open interoperability protocol.
Google donates the project to neutral governance under the Linux Foundation.
The protobuf becomes normative; signed cards, multi-tenancy and three standard bindings land.
The maintenance release aligns media types, task states and error transcoding.
a2a.proto is the universal source of truth; JSON follows ProtoJSON naming and enum rules.
Agent Cards can be signed using JCS canonicalization and JWS, with key material resolved out of band.
A2A-Version and A2A-Extensions make compatibility and opt-in behavior visible on the wire.
Tasks and requests can carry tenant context while authorization remains an application responsibility.
mTLS and OAuth device-code support join API keys, bearer tokens and familiar OAuth flows.
ListTasks and the canonical CamelCase operation names round out task management.
kind discriminators, uses ROLE_USER / ROLE_AGENT and TASK_STATE_* enums, and moves protocol choices into supportedInterfaces[]. During migration, dual 0.3 + 1.0 support is the safest compatibility path.Start from maintained protocol types instead of hand-rolling every wire object.
Interoperability expands the trust boundary. A compliant wire format is not a trust decision: authenticate identity, authorize each action and validate every returned value.
Use HTTPS in production; TLS 1.3 is recommended. Validate server certificates and verify signed Agent Cards before trusting discovered metadata.
The card declares acceptable schemes; credentials are acquired separately. Never place secrets in messages, parts, task metadata or Agent Cards.
Check tenant, organization, task and skill access on every operation. Possessing a task ID must never grant access by itself.
Sanitize card text, messages, URLs, files and structured data before rendering or feeding it to a model. Defend against prompt injection and unsafe URLs.
Atlas by Humaps exposes verifiable public-intelligence research as an A2A v1.0 service, with v0.3 compatibility for existing clients.
curl -X POST https://api.civensia.com/api/a2a \
-H "A2A-Version: 1.0" \
-H "Content-Type: application/a2a+json" \
-H "X-Atlas-API-Key: $ATLAS_API_KEY" \
-d '{
"jsonrpc": "2.0",
"id": 1,
"method": "SendMessage",
"params": {
"message": {
"messageId": "message-1",
"role": "ROLE_USER",
"parts": [
{
"text": "Compare population growth in Amsterdam and Rotterdam since 2015, using official sources."
}
]
}
}
}'Skill IDs remain stable while connectors evolve. Use the detail endpoint when your client needs the current tools and input schemas.
query.statistics.nlCBS StatLine and natural-language statistical research
query.statistics.euEurostat and member-state comparison
query.statistics.worldWorld Bank, UN, WHO and other public sources
query.public_data.countryRuntime discovery across national open-data systems
query.policy.nlParliamentary and official publications
query.policy.euEUR-Lex and European parliamentary material
query.official_documents.worldWorldwide government-document discovery
query.geo.euGeocoding, boundaries and cartography
query.geo.worldWorldwide geocoding and administrative areas
query.geospatial.layers.worldDiscover and query public map layers
query.events.liveConflict, governance and disaster signals
query.research.scholarlyOpenAlex and multi-source research synthesis
analysis.statisticsTests, trends, regression and interpretation
analysis.spatialChoropleths, Moran’s I and hot-spot analysis
briefing.policyStructured, sourced DOCX and PPTX output
research.documentVersioned long-form research and patches
This is a Humaps extension, not part of the core A2A specification. Pass a stable skill ID and choose a mode, or omit both and let Atlas route the natural-language task.
Strict schema, one declared connector, deterministic response. Available on supported skills and free within direct-mode limits.
Natural language, planning and multi-tool synthesis. Plan-gated and metered by usage units.
The protocol is intentionally small. The hard production questions live at the edges: trust, lifecycle, compatibility and choosing the right boundary.
No. MCP connects an AI application to tools, resources and context. A2A connects one autonomous agent to another agent and gives that collaboration a discoverable identity, task lifecycle and delivery model. A common architecture uses both.
This guide was reviewed against the official specification, project documentation and release history on 2 August 2026. A2A evolves quickly; use the normative protobuf and current specification for implementation decisions.
Normative operations, data model, security and bindings
Official migration and compatibility overview
Agent Cards, messages, tasks, parts and artifacts
State transitions and multi-turn interactions
Well-known cards, registries and direct configuration
Python, JavaScript, Java, Go, .NET and Rust
Version history and v1.0.1 release notes
TCK, Inspector, extensions and interoperability work
Linux Foundation ecosystem and production-adoption update
Google’s guide to A2A, MCP, UCP, A2UI and AG-UI