multi-agent-handoff-architect

Designs versioned JSON Schema contracts for inter-agent handoffs with binary ownership and cross-boundary observability.

Updated Apr 26, 2026
One-click install
npx skills add https://github.com/Jin9/skillify-foundation --skill multi-agent-handoff-architect-jin9
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: multi-agent-handoff-architect
Source: https://github.com/Jin9/skillify-foundation/tree/main/treasury/multi-agent-handoff-architect
Command: npx skills add https://github.com/Jin9/skillify-foundation --skill multi-agent-handoff-architect-jin9

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Unstructured, free-text handoffs between AI agents cause context loss, state corruption, and cascading errors in multi-agent systems. This Skill turns inter-agent handoffs into versioned, schema-validated API contracts so control, context, and responsibility transfer reliably between agents. ## Core Features & Use Cases - Handoff contract design: Emits a JSON Schema (draft 2020-12) with six required fields — taskId, intent, state, confidence, provenance, schemaVersion — plus additionalProperties: false and no free-text fields. - Topology decision record: Applies a four-step gate (workflow vs single agent vs multi-agent squad) defaulting to orchestrator-worker, with a documented rationale. - Binary write-ownership mapping: Assigns exactly one writer per shared-state element with explicit ownership-transfer points, plus autonomy levels calibrated to task reversibility. - Use Case: When defining the payload between a planner agent and a coder agent, use this Skill to produce the handoff design doc, the topology decision record, and a validated JSON Schema with correlation IDs for cross-agent tracing. ## Quick Start Ask the agent to design a handoff contract between your planner and coder agents, including the JSON Schema payload and a topology decision record.

Frequently Asked Questions about multi-agent-handoff-architect

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I design a handoff contract between AI agents?▼

Define the handoff as a JSON Schema with six required fields: taskId, intent, state, confidence, provenance, and schemaVersion. Validate each payload at generation time with repair-on-failure, and forbid free-text notes fields so context transfers as structured data.

When should I use multi-agent instead of a single agent?▼

Default to a single agent unless the task is parallelizable, read-heavy, has independent sub-problems, and includes a hard verifier that rejects rather than blends outputs. If the task is a deterministic flowchart covering 90%+ of cases, use a workflow instead of agents.

What fields should an inter-agent handoff payload include?▼

A handoff payload requires taskId, an enumerated intent, a structured state object (200-500 tokens), a confidence score from 0 to 1, provenance with traceId and spanId, and a semver schemaVersion. Free-text narrative fields are forbidden.

Why do free-text handoffs between agents fail?▼

Free-text narrative notes are the dominant source of context loss in multi-agent systems, causing cascading errors and state corruption. Typed, schema-validated contracts with additionalProperties set to false prevent this drift.

Does this skill build the agents or telemetry pipeline?▼

No. It produces only the handoff design doc, topology decision record, and JSON Schema. Telemetry instrumentation, agent permission governance, AGENTS.md prose, and pipeline stage composition are delegated to sibling skills.