Crate a2a_protocol_server 

A2A protocol v1.0 — server framework.

Provides RequestHandler and AgentExecutor for implementing A2A agents over HTTP/1.1 and HTTP/2 using hyper 1.x.

Quick start

1. Implement AgentExecutor with your agent logic.
2. Build a RequestHandler via RequestHandlerBuilder.
3. Wire JsonRpcDispatcher or RestDispatcher into your hyper server.

Module overview

Module	Contents
error	ServerError, ServerResult
executor	AgentExecutor trait
executor_helpers	boxed_future, agent_executor! macro
handler	RequestHandler, SendMessageResult, HandlerLimits
builder	RequestHandlerBuilder
store	TaskStore, InMemoryTaskStore, SqliteTaskStore (sqlite feature)
streaming	Event queues, SSE response builder
push	Push config store, push sender
agent_card	Static/dynamic agent card handlers
serve	serve(), serve_with_addr, Dispatcher
dispatch	JsonRpcDispatcher, RestDispatcher
interceptor	ServerInterceptor, ServerInterceptorChain
rate_limit	RateLimitInterceptor, RateLimitConfig
request_context	RequestContext
call_context	CallContext (includes HTTP headers for auth)
metrics	Metrics trait (request counts, latency, errors)
tenant_resolver	TenantResolver, HeaderTenantResolver, BearerTokenTenantResolver, PathSegmentTenantResolver
tenant_config	PerTenantConfig, TenantLimits
otel	OtelMetrics, OtelMetricsBuilder, init_otlp_pipeline (otel feature)

Axum integration

Enable the axum feature flag to use A2aRouter for idiomatic Axum integration. See the dispatch::axum_adapter module for details.

gRPC transport

Enable the grpc feature flag to use GrpcDispatcher for gRPC transport (tonic-backed). See the dispatch::grpc module for details.

Rate limiting

Built-in rate limiting is available via RateLimitInterceptor, a fixed-window per-caller interceptor. For advanced use cases (sliding windows, distributed counters), use a reverse proxy (nginx, Envoy) or a custom ServerInterceptor.

Re-exports

pub use agent_card::AgentCardProducer;

pub use agent_card::DynamicAgentCardHandler;

pub use agent_card::HotReloadAgentCardHandler;

pub use agent_card::StaticAgentCardHandler;

pub use agent_card::CORS_ALLOW_ALL;

pub use builder::RequestHandlerBuilder;

pub use call_context::CallContext;

pub use dispatch::axum_adapter::A2aRouter;

pub use dispatch::WebSocketDispatcher;

pub use dispatch::CorsConfig;

pub use dispatch::DispatchConfig;

pub use dispatch::JsonRpcDispatcher;

pub use dispatch::RestDispatcher;

pub use dispatch::GrpcConfig;

pub use dispatch::GrpcDispatcher;

pub use error::ServerError;

pub use error::ServerResult;

pub use executor::AgentExecutor;

pub use executor_helpers::boxed_future;

pub use executor_helpers::EventEmitter;

pub use handler::HandlerLimits;

pub use handler::RequestHandler;

pub use handler::SendMessageResult;

pub use interceptor::ServerInterceptor;

pub use interceptor::ServerInterceptorChain;

pub use metrics::ConnectionPoolStats;

pub use metrics::Metrics;

pub use otel::OtelMetrics;

pub use push::HttpPushSender;

pub use push::InMemoryPushConfigStore;

pub use push::PushConfigStore;

pub use push::PushRetryPolicy;

pub use push::PushSender;

pub use push::TenantAwareInMemoryPushConfigStore;

pub use rate_limit::RateLimitConfig;

pub use rate_limit::RateLimitInterceptor;

pub use request_context::RequestContext;

pub use serve::serve;

pub use serve::serve_with_addr;

pub use serve::Dispatcher;

pub use store::InMemoryTaskStore;

pub use store::TaskStore;

pub use store::TaskStoreConfig;

pub use store::TenantAwareInMemoryTaskStore;

pub use store::TenantContext;

pub use store::TenantStoreConfig;

pub use push::SqlitePushConfigStore;

pub use push::TenantAwareSqlitePushConfigStore;

pub use store::Migration;

pub use store::MigrationRunner;

pub use store::SqliteTaskStore;

pub use store::TenantAwareSqliteTaskStore;

pub use push::PostgresPushConfigStore;

pub use push::TenantAwarePostgresPushConfigStore;

pub use store::PgMigration;

pub use store::PgMigrationRunner;

pub use store::PostgresTaskStore;

pub use store::TenantAwarePostgresTaskStore;

pub use streaming::EventQueueManager;

pub use streaming::EventQueueReader;

pub use streaming::EventQueueWriter;

pub use streaming::InMemoryQueueReader;

pub use streaming::InMemoryQueueWriter;

pub use tenant_config::PerTenantConfig;

pub use tenant_config::TenantLimits;

pub use tenant_resolver::BearerTokenTenantResolver;

pub use tenant_resolver::HeaderTenantResolver;

pub use tenant_resolver::PathSegmentTenantResolver;

pub use tenant_resolver::TenantResolver;

Modules

agent_card

Agent card HTTP handlers (static, dynamic, and caching utilities).

builder

Builder for RequestHandler.

call_context

Call context for server-side interceptors.

dispatch

HTTP dispatch layer — JSON-RPC and REST routing.

error

Server-specific error types.

executor

Agent executor trait.

executor_helpers

Ergonomic helpers for implementing AgentExecutor.

handler

Core request handler — protocol logic layer.

interceptor

Server-side interceptor chain.

metrics

Metrics hooks for observing handler activity.

otel

OpenTelemetry integration for the A2A server.

push

Push notification configuration storage and delivery.

rate_limit

Token-bucket rate limiter as a ServerInterceptor.

request_context

Request context passed to the AgentExecutor.

serve

Server startup helpers.

store

Task storage backend.

streaming

Streaming infrastructure for SSE responses and event queues.

tenant_config

Per-tenant configuration for multi-tenant A2A servers.

tenant_resolver

Tenant resolution for multi-tenant A2A servers.

Macros

agent_executor

Generates an AgentExecutor implementation from a closure-like syntax.