A2A Php Sdk Laravel Package

Product Decisions This Supports

• Build vs. buy for agent interoperability: A2A (Agent-to-Agent) protocol support is infrastructure, not differentiation. Adopting a 1.0.0 SDK lets the team ship agent-to-agent workflows without owning protocol conformance, transport negotiation, or protobuf schema maintenance.
• Feature feasibility for an "agent ecosystem" roadmap item: This package makes it realistic to commit to cross-vendor agent delegation, task handoff, and long-running task subscriptions on the near-term roadmap, since all 11 protocol operations are covered across JSON-RPC, REST, and optional gRPC.
• Architecture choice: how agents integrate. The SDK's transport flexibility (HTTP-only baseline, gRPC as an opt-in accelerator) directly informs the decision to standardize on HTTP for v1 and defer the gRPC/OpenSwoole operational investment until volume justifies it.
• Multi-tenant / enterprise readiness scoping: Principal and tenant isolation, persistent tasks, worker leases, and webhook retries are included rather than bespoke. That converts "enterprise agent isolation" from a custom build to a configuration and validation exercise.
• Operability commitments: PSR-3 logging and Prometheus metrics out of the box mean we can promise observability SLAs to platform stakeholders without writing instrumentation.
• Vendor-risk positioning: Because the protocol is the contract rather than the vendor, adopting A2A is a hedge — we retain the ability to swap agent providers while keeping our integration layer stable.

When to Consider This Package

Adopt when:

• The roadmap includes agent-to-agent or cross-organization agent delegation and we want a conformant client/server rather than a homegrown protocol implementation.
• We are on, or willing to move to, PHP 8.4+ — this is a hard floor, not a soft preference. Confirm the Laravel version in use supports 8.4 before committing.
• We need long-running tasks with resumable state (subscriptions, cancellation, leases, retries) rather than simple request/response calls.
• We need multi-tenant isolation or per-principal authorization at the protocol layer.
• We want a low-friction HTTP-first deployment — the package explicitly does not require native gRPC extensions or OpenSwoole, which materially reduces platform review and deployment risk.

Look elsewhere or defer when:

• Formal TCK certification is a contractual or procurement requirement. This release explicitly does not claim full official TCK certification: 244 passed / 3 failed / 18 skipped against the pinned suite, with all three failures tied to a documented upstream defect (CORE-SEND-003). With a local diagnostic correction it reaches 247/0/18, but that is not the same as certification. Treat this as "validated against a known upstream bug," not "certified."
• The organization is not ready to version-lock to a 1.0.0 from a single-maintainer package. This is a first release; ecosystem maturity, issue-response track record, and long-term maintenance signals are unproven regardless of the strong CI story.
• We only need basic agent invocation — a thin HTTP client may be less surface area than a full protocol SDK.
• We need mature Laravel-native ergonomics (service provider autodiscovery, publishable configs, facades, queue integration). The release notes describe a PHP SDK, not a Laravel-idiomatic wrapper; expect integration glue work.

How to Pitch It (Stakeholders)

To executives: This is a free, standards-based SDK that lets us plug into the emerging agent-to-agent ecosystem without a custom protocol build. It covers the full A2A 1.0.0 surface — all 11 operations, streaming, persistence, tenant isolation, metrics — and it deploys over plain HTTP, so no new infrastructure or specialized runtime is required. The main caveat is maturity: it is a 1.0.0 from a small project and is not formally TCK-certified, so we should position early adoption as a pilot behind a feature flag rather than a platform-wide commitment.

To engineering: Treat this as a well-engineered but young dependency. The release has credible verification — CI covering unit and network tests, TLS/mTLS, PHPStan, style, generated contract verification, dependency audit, and a clean-consumer install — plus a published ZIP with SHA256SUMS. Streaming (SSE, optional gRPC), typed protobuf messages, validation, transport error mapping, worker leases, and webhook retries are all in scope. Note hard requirement PHP 8.4+, no native gRPC or OpenSwoole needed for HTTP, and the TCK baseline is 244/3/18 with the three failures attributable to a documented upstream defect. Plan for a Laravel integration layer (config, service provider wiring, queue bridging) since this is a framework-agnostic SDK.

Recommendation: Proceed to a scoped pilot with a dependency-pinning and upgrade-review policy, and revisit for broader adoption once TCK status or maintainer track record improves.