# Agent2Agent Protocol > The A2A (Agent2Agent) Protocol community site: documentation and developer guides, agent collaboration tutorials, technical topics, SDKs, tools, and implementation experience. ## A2A Protocol Technical Documentation - [A2A Protocol Technical Documentation](https://agent2agent.info/docs/): Technical documentation for the A2A Protocol (Agent2Agent): introduction, core concepts, task lifecycle, streaming, agent discovery, enterprise features, a Python tutorial, and official SDKs, aligned with A2A v1.0.1. - [A2A Protocol Explained: Agent2Agent Concepts and Architecture](https://agent2agent.info/docs/introduction/): An introduction to the A2A (Agent2Agent) Protocol: background and positioning, design principles, participants, core concepts, and the three protocol bindings, plus how it relates to MCP and where it applies. - [A2A Python Tutorial: Build, Run and Call an Agent](https://agent2agent.info/docs/tutorials/python/): The A2A Python tutorial: build, run, and call an A2A agent with the Python SDK, covering setup, skills and the Agent Card, the executor, the server, the client, streaming, and multi-turn interactions. - [3. Agent Skills and Agent Card](https://agent2agent.info/docs/tutorials/python/3-agent-skills-and-card/): Before an A2A agent can do anything, it defines what it can do (AgentSkill) and how it is discovered (AgentCard). The official helloworld sample shows the key attributes and code. - [4. The Agent Executor](https://agent2agent.info/docs/tutorials/python/4-agent-executor/): The AgentExecutor is how an A2A agent handles requests and emits events: the execute and cancel methods, and the Helloworld flow from task creation to completion. - [5. Starting the Server](https://agent2agent.info/docs/tutorials/python/5-start-server/): With an Agent Card and an executor in place, set up and start the A2A server: run the Helloworld agent with DefaultRequestHandler, create_agent_card_routes, create_jsonrpc_routes, Starlette, and Uvicorn. - [6. Interacting with the Server](https://agent2agent.info/docs/tutorials/python/6-interact-with-server/): With the Helloworld server running, use the official test_client.py: fetch the Agent Card with A2ACardResolver, send non-streaming and streaming messages, and read the returned Task and events. - [7. Streaming and Multi-Turn Interactions (LangGraph Example)](https://agent2agent.info/docs/tutorials/python/7-streaming-and-multiturn/): Explore advanced A2A features with the official LangGraph Currency Agent: LLM integration, task state management, status and artifact streaming events, and TASK_STATE_INPUT_REQUIRED multi-turn conversations. - [8. Next Steps](https://agent2agent.info/docs/tutorials/python/8-next-steps/): What to do after the A2A Python tutorial: review what you learned, then go deeper with more samples, the protocol documentation, key topics, your own agent, advanced features, and the community. - [A2A Implementation Guides](https://agent2agent.info/docs/guides/): Task-based implementation guides for the A2A Protocol (Agent2Agent) v1.0.1: publish an Agent Card, run and recover tasks, implement security, harden push notifications, make tasks idempotent, and prepare a production release. - [A2A Agent Card Implementation Guide: Build, Validate, Sign](https://agent2agent.info/docs/guides/agent-card-implementation/): Publish an A2A v1.0.1 Agent Card that clients can discover and trust: declare interfaces and capabilities, describe skills and security, validate the card, optionally sign it, and map runtime errors back to card fixes. - [A2A Task Runtime Guide: Run, Stream, Resume, Recover](https://agent2agent.info/docs/guides/task-runtime/): Run an A2A v1.0.1 task end to end: make your first exchange from a fetched Agent Card, choose Message or Task and the delivery mode, learn which operation each state accepts, check an event stream, recover from broken streams, and migrate from v0.3 without the blocking-default trap. - [A2A Production Readiness Checklist: Test, Observe, Roll Back](https://agent2agent.info/docs/guides/production-readiness/): Take an A2A v1.0.1 server from working in development to a defensible production release: run a discovery preflight, test authentication and task isolation, verify streaming and push behavior, define observability evidence, and prepare a capability-safe rollback. - [A2A Security Implementation: Authentication, Authorization, and In-Task Auth](https://agent2agent.info/docs/guides/security-implementation/): Turn the A2A v1.0.1 security requirements into server code: declare security schemes, answer authentication on all three protocol bindings, scope every operation to its caller, implement TASK_STATE_AUTH_REQUIRED, and self-test with negative probes. - [A2A Push Notification Security: Sender Validation and Receiver Hardening](https://agent2agent.info/docs/guides/push-receiver-hardening/): Destination validation, source authentication, durable receipt, and duplicate handling for A2A v1.0.1 push notifications. Fault probes require configured tasks, credentials, and a controlled test environment; runtime assertions remain unverified. - [A2A Task Idempotency and Recovery: Duplicates, Dropped Streams, and Stalled Tasks](https://agent2agent.info/docs/guides/task-idempotency-recovery/): Implement reliable task handling on A2A v1.0.1: handle duplicate messages with messageId and contextId, recover from dropped streams with SubscribeToTask and GetTask, detect and terminate tasks stalled in WORKING, and give accepted tasks a definite outcome after a restart. A runnable reference endpoint verifies duplicates, crashes, restarts, and reconnections by counting business executions. - [A2A and MCP in Practice: Agent Task Handoff and Recovery](https://agent2agent.info/docs/guides/mcp-tool-task-handoff/): How A2A and MCP work together in a followable A2A v1.0.1 scenario: a coordinator agent discovers a ticket-analysis specialist, starts, supplements, and tracks the task in one context, receives Messages and Artifacts, and recovers from timeouts, dropped streams, and required authorizations. How the specialist reads data and calls tools over MCP internally is another page's boundary. - [Core Topics](https://agent2agent.info/docs/topics/): Understand the key topics of the A2A protocol, from fundamentals to production practice: core concepts, task lifecycle, streaming, agent discovery, enterprise features, multi-tenancy, extensions, and custom protocol bindings. - [What Is the A2A Protocol](https://agent2agent.info/docs/topics/what-is-a2a/): A2A (Agent2Agent) is an open standard for communication and collaboration between AI agents. Learn the problems it solves, its design principles, its relationship to MCP and frameworks, and a request's lifecycle. - [A2A Key Concepts and Components](https://agent2agent.info/docs/topics/key-concepts/): A2A defines how agents interact with a small set of core concepts: actors, Agent Card, Task, Message, Part, and Artifact, plus three interaction mechanisms: request/response, streaming, and push notifications. - [Streaming and Asynchronous Operations](https://agent2agent.info/docs/topics/streaming-and-async/): A2A offers two mechanisms for long-running tasks: real-time streaming over SSE and push notifications for disconnected clients. Learn the capability flags, event structure, use cases, and push notification security considerations. - [Agent Discovery in A2A](https://agent2agent.info/docs/topics/agent-discovery/): Learn the role of the Agent Card, three discovery strategies (well-known URI, registries, direct configuration), how to secure and sign Agent Cards, and HTTP caching guidance. - [Enterprise Implementation of A2A](https://agent2agent.info/docs/topics/enterprise-ready/): A2A reuses existing TLS, OAuth2/OpenID Connect, distributed tracing, and API management. Covers transport security, authentication, authorization, data privacy, observability, and API governance. - [Extensions in A2A](https://agent2agent.info/docs/topics/extensions/): Extensions layer new capabilities onto A2A: data-only, profile, method, and state machine extensions. Learn declaration, required extensions, activation via the A2A-Extensions header, versioning, and security. - [Core Objects](https://agent2agent.info/docs/concepts/): A2A v1.0.1 object reference: Agent Card, Task, Message, Part, and Artifact, with fields, protocol rules, and request and response examples from the official documentation. - [A2A Agent Card: Structure, Discovery, and Signing](https://agent2agent.info/docs/concepts/agentcard/): The Agent Card is the JSON document an A2A server publishes, usually at /.well-known/agent-card.json. Required fields, supportedInterfaces, capabilities, skills, security schemes, JWS signatures, and caching, with examples from the A2A v1.0 specification. - [A2A Task: Lifecycle, States, and Operations](https://agent2agent.info/docs/concepts/task/): A Task is the core unit of work in the A2A Protocol: a stateful object with an ID, a status, artifacts, and history. Task fields, the eight TASK_STATE_* values, terminal and interrupted states, multi-turn input, GetTask, ListTasks, CancelTask, and update delivery, with examples from the A2A v1.0 specification. - [A2A Message: Roles, Fields, and When to Use It](https://agent2agent.info/docs/concepts/message/): A Message is one unit of communication between an A2A client and an agent. Message fields, ROLE_USER and ROLE_AGENT, how messageId, contextId, taskId, and referenceTaskIds relate, what messages are for versus artifacts, and examples from the A2A v1.0 specification. - [A2A Part: Text, File, and Structured Data Content](https://agent2agent.info/docs/concepts/part/): A Part is the content container inside A2A Messages and Artifacts. It holds exactly one of text, raw bytes, a URL, or structured data, plus mediaType, filename, and metadata. Examples for text, file upload and download, and structured data, with the v0.3 to v1.0 changes. - [A2A Artifact: Task Outputs and Streaming Chunks](https://agent2agent.info/docs/concepts/artifact/): An Artifact is a concrete output of an A2A task: a document, an image, or structured data. Artifact fields, how artifacts differ from messages, streaming with TaskArtifactUpdateEvent append and lastChunk, and refining an artifact in a follow-up task, with examples from the A2A v1.0 specification. - [A2A AgentSkill: Describing What an Agent Can Do](https://agent2agent.info/docs/concepts/agentskill/): AgentSkill is one entry in the skills list of an A2A Agent Card. Required fields, per-skill input and output modes, per-skill security requirements, a real example from the A2A v1.0 specification, and what changed since v0.3. - [A2A SecurityScheme: Declaring How Clients Authenticate](https://agent2agent.info/docs/concepts/securityscheme/): How an A2A Agent Card declares authentication: the five SecurityScheme types, OAuth 2.0 flows (including Device Code and PKCE), SecurityRequirement, and the naming pitfalls between the proto, the specification, and the v1.0 migration guide. - [A2A Reference](https://agent2agent.info/docs/reference/): Lookup pages for the A2A Protocol: error codes across the JSON-RPC, gRPC, and HTTP bindings, Agent Card proto field numbers, and interoperability across bindings and versions (v0.3.0 to v1.0.1). - [A2A Error Codes: JSON-RPC, gRPC, and HTTP Mapping](https://agent2agent.info/docs/reference/errors/): Lookup table for the nine A2A-specific errors: JSON-RPC code, gRPC status, HTTP status, and ErrorInfo reason. Which operation raises which error, the standard JSON-RPC codes, and the same error as it appears in each binding. A2A v1.0.1. - [A2A Agent Card Proto Field Numbers: v0.3.0 vs v1.0.1](https://agent2agent.info/docs/reference/agentcard-field-numbers/): Field-by-field comparison of AgentCard, AgentCapabilities, AgentInterface, and AgentSkill in a2a.proto at v0.3.0 and v1.0.1: reused numbers, renamed fields, and where the extended Agent Card flag moved (AgentCard field 13 to AgentCapabilities field 4). - [A2A Interoperability Reference: Bindings, Versions, Wire Differences](https://agent2agent.info/docs/reference/interoperability/): Lookup for when two A2A implementations do not connect or must support v0.3 and v1.0 side by side: interface selection, A2A-Version handling, the 11 operations across JSON-RPC, gRPC, and HTTP+JSON, and every wire-level change from v0.3.0 to v1.0.1. - [Official SDKs](https://agent2agent.info/docs/sdk/): A guide to the official A2A SDKs and sample repositories: Python, Go, Java, JavaScript, C#/.NET, and Rust, plus companion tools such as the A2A CLI, the Inspector, and the TCK. - [Community Hub](https://agent2agent.info/docs/community/): The A2A community entry point: Discord and meetings, official announcements, agent frameworks with built-in A2A support, community SDKs, plus partners, the roadmap, and the v1.0 migration guide. - [A2A Protocol FAQ](https://agent2agent.info/docs/faq/): Answers to common A2A Protocol questions: MCP, agent discovery, bindings, multi-tenancy, Message vs Task, multi-turn interaction, streaming, authentication, versioning, SDKs, governance, and roadmap. ## Optional - [Privacy Policy](https://agent2agent.info/privacy/): How this A2A Protocol reference site handles data: hosting logs, usage statistics, advertising, browser storage, third-party links, and how to contact us about your data. - [Contact](https://agent2agent.info/contact/): Contact the A2A community to share agent collaboration experience and resources, suggest tutorials, report problems, or discuss collaboration and privacy. - [About the A2A Community: Tutorials, Documentation, and Discussion](https://agent2agent.info/about/): An A2A Protocol documentation and developer community with tutorials, agent collaboration topics, SDKs, tools, and practical implementation experience.