Result JSON Schema for AI Agent Workflows
A schema pattern for turning open-ended agent work into a product-safe API response.
Audience: Backend engineers integrating agent runs into production systems.
The problem
Without a result schema, every integration becomes prompt parsing and fragile string matching.
Implementation path
Define schema_version, summary, body.type, body.data, and artifacts. Let each skill own body.data while the runtime enforces the outer contract.
Tradeoffs and failure modes
The outer schema should be strict; the inner body can stay skill-specific.
Base schema
{ "schema_version": "argo.result.v1", "summary": "...", "body": { "type": "...", "data": {} }, "artifacts": [] }
Run this on Argo