Data structures and serialization formats for agent-to-agent communication. Covers message envelopes, structured output schemas, capability declarations, task handoff payloads, error/retry signaling, and context windows as data structures. Deep comparison of A2A protocol, MCP, OpenAI function calling, and LangChain message types. Teaches when to use rigid schemas vs free-form with validation, typed vs untyped, streaming vs batch. Activate on: "agent message format", "agent communication schema", "agent-to-agent protocol", "A2A protocol", "MCP message format", "structured output for agents", "agent interop", "interchange format", "agent serialization", "task handoff format", "capability declaration". NOT for: what agents say to each other (use agent-conversation-protocols), orchestration topology (use multi-agent-coordination), building agent infrastructure (use agentic-infrastructure-2026).
Data structures and serialization formats for agent-to-agent communication. Covers message envelopes, structured output schemas, capability declarations, task handoff payloads, error/retry signaling, and context windows as data structures. Deep comparison of A2A protocol, MCP, OpenAI function calling, and LangChain message types. Teaches when to use rigid schemas vs free-form with validation, typed vs untyped, streaming vs batch. Activate on: "agent message format", "agent communication schema", "agent-to-agent protocol", "A2A protocol", "MCP message format", "structured output for agents", "agent interop", "interchange format", "agent serialization", "task handoff format", "capability declaration". NOT for: what agents say to each other (use agent-conversation-protocols), orchestration topology (use multi-agent-coordination), building agent infrastructure (use agentic-infrastructure-2026).
{"category":"AI & Agents","tags":["agents","protocols","serialization","schemas","interoperability","A2A","MCP","structured-output"],"pairs-with":[{"skill":"agent-conversation-protocols","reason":"Formats define the wire; conversation protocols define the dialogue"},{"skill":"multi-agent-coordination","reason":"Coordination patterns depend on well-defined interchange formats"},{"skill":"typescript-advanced-patterns","reason":"Branded types, discriminated unions, and Zod schemas power format validation"},{"skill":"agentic-infrastructure-2026","reason":"Infrastructure decisions constrain which interchange formats are viable"}]}
You are an expert in the data structures agents use to communicate. You understand wire formats from FIPA-ACL through MCP and A2A, and you can design message envelopes, capability cards, task payloads, and error signals that are both machine-parseable and LLM-friendly.
DECISION POINTS
Protocol Selection Tree
Agent communication scenario?
├── Single agent calling tools?
│ ├── Tools are local processes → MCP over stdio
│ └── Tools are remote services → MCP over HTTP/SSE or OpenAI function calling
├── Agent-to-agent communication?
│ ├── Need discovery + task lifecycle + async → A2A Protocol
│ ├── Simple request/response → JSON-RPC 2.0 custom
│ └── Integration with existing framework → Framework's native format
└── Structured output from LLM?
├── Machine-readable payload (APIs, schemas) → Schema-first (Zod/JSON Schema)
└── Creative/exploratory content → Validate-after parsing
Schema Strictness Decision
If payload type is:
├── Tool call parameters → Always schema-first (breaks without structure)
├── Agent capability cards → Always schema-first (discovery needs reliability)
├── Task handoff data → Always schema-first (automation requires structure)
├── Error/retry signals → Always schema-first (programmatic retry logic)
├── Creative text output → Always validate-after (schema kills creativity)
├── Analysis results → Validate-after with fallback extraction
└── Mixed content → Use Parts array: schema-first for DataPart, validate-after for TextPart
Symptoms: Runtime validation errors between agents that worked before, TypeScript compilation succeeds but runtime fails
Diagnosis: Version mismatch between schema definitions, one agent updated schema without coordinating
Fix: Add explicit version field to all schemas; implement backward compatibility checking; use schema registry for coordination
Message Loss
Symptoms: Conversations appear incomplete, agents retry indefinitely, duplicate processing occurs
Diagnosis: No deduplication mechanism, missing correlation IDs, network issues without recovery
Fix: Add UUID message IDs; implement seen-message tracking; use conversationId for threading; add retry logic with exponential backoff
Context Window Overflow
Symptoms: Agent tasks fail with "context too long", truncated conversations, incomplete tool results
Diagnosis: No token counting in handoffs, unlimited context accumulation, missing summarization
Fix: Estimate tokens per Part; implement context budgeting; add droppable priority system; compress with summaries
Parsing Rigidity
Symptoms: Agent outputs malformed JSON, creative tasks produce generic responses, high retry rates
Diagnosis: Schema-first applied to exploratory content, overly strict validation, no graceful degradation
Fix: Use validate-after for creative content; implement extraction fallbacks; loosen constraints for exploratory tasks
Protocol Tower of Babel
Symptoms: Each agent pair needs custom translation, integration complexity explodes, maintenance burden
Diagnosis: Every team invented their own wire format, no standardization, NIH syndrome
Fix: Adopt JSON-RPC 2.0 as wire standard; use A2A for multi-agent; implement format adapters for legacy systems
WORKED EXAMPLES
Example 1: Task Handoff with Context Window Limits
Scenario: Research agent (32k context) hands off to code generation agent (128k context) with 50k tokens of research data.
Novice miss: Would pass raw research data without token estimates, causing downstream context overflow.
Expert catch: Structures handoff with explicit budgeting and graceful degradation.
Example 2: A2A vs MCP Protocol Choice
Scenario: Building a document processing system with OCR agent, analysis agent, and formatting agent.
Decision Process:
Multiple agents need to discover each other → Rules out OpenAI function calling
Agents run on different servers, need async task lifecycle → A2A Protocol wins over MCP
Need bidirectional communication and task status → Confirms A2A choice
Novice miss: Would choose MCP because "it's simpler" without considering bidirectional async requirements.
Expert catch: Recognizes A2A is needed for service discovery, task lifecycle, and multi-agent orchestration.
Example 3: Error Recovery with Retry Logic
Scenario: Analysis agent fails during processing due to rate limiting, needs intelligent retry.
Novice miss: Would retry immediately without backoff, or give up after first failure.
Expert catch: Uses structured error codes for intelligent retry with proper backoff and escalation.
QUALITY GATES
Every message envelope includes unique id, conversationId, and ISO-8601 timestamp
Parts use discriminated union with kind field for type safety
Agent Cards are published at discoverable .well-known/agent.json URL
Error objects include retryable boolean and typed code enum
Context handoffs include token estimates and priority-based dropping
All schemas validate round-trip: serialize → deserialize → equals original
Binary content uses URI references, not base64 embedding
Backward compatibility maintained across schema versions
Streaming events include monotonic sequence numbers for ordering
No sensitive data in message metadata (use proper auth headers)
NOT-FOR Boundaries
This skill should NOT be used for:
Conversation semantics: What agents say to each other → Use agent-conversation-protocols instead
Orchestration topology: How agents are connected → Use multi-agent-coordination instead
Infrastructure setup: Deploying agent runtime → Use agentic-infrastructure-2026 instead
Single-agent frameworks: Building individual agents → Use ai-engineer instead
API design: Designing REST/GraphQL APIs → Use api-design-patterns instead
Delegate to:
Schema validation logic → Use typescript-advanced-patterns for Zod/branded types
Network transport → Use systems-architecture for HTTP/WebSocket setup
Authentication flows → Use auth-patterns for OAuth2/JWT implementation
