[plan] Address schema field dependency and relationship gaps

[github-actions]

Overview

This tracking issue addresses the 7 critical relationship gaps identified in the graph-based field dependency analysis. These are schema-documentation gaps where the schema permits invalid configurations that the compiler correctly rejects.

Source: Discussion #7569

Problem Statement

The JSON schema currently describes field relationships only in prose within description fields, and doesn't leverage JSON Schema's built-in relationship features (oneOf, if/then, dependencies, etc.). This creates several issues:

• Schema validation passes for mutually exclusive field combinations
• Engine-specific feature compatibility is undocumented
• Conditional requirements aren't enforced at schema level
• Users discover validation errors only at compile time

Planned Tasks

This work is broken into 5 focused sub-issues:

1. Add mutual exclusivity constraints - Implement oneOf constraints for field pairs that cannot coexist
2. Document engine-specific features - Add compatibility metadata to schema for engine-locked features
3. Add conditional validation - Implement if/then constraints for conditional requirements
4. Add array constraints - Add minItems where fields require at least one value
5. Refactor to use dependencies keyword - Migrate relationship prose to JSON Schema standard features

Success Criteria

• Schema validation catches mutually exclusive field combinations
• Engine compatibility is documented in schema metadata
• Conditional requirements are schema-enforced
• All array fields have appropriate minItems constraints
• Field relationships use JSON Schema standard keywords

References

• Analysis: Discussion Graph-Based Field Dependency Analysis - 7 Critical Relationship Gaps Found #7569
• Schema Location: pkg/parser/schemas/
• Validation Code: pkg/workflow/compiler_filters_validation.go, pkg/workflow/compiler_safe_outputs.go, pkg/workflow/mcp_config_validation.go

AI generated by Plan Command for discussion #7569