What problem does it solve?
AI agents generating workflows in cc-wf-studio sometimes pick the wrong node types or follow false constraints, and the root cause is usually biased text inside resources/workflow-schema.json rather than missing rules. This Skill guides you through diagnosing and fixing those biases by treating schema edits as prompt engineering, not type definitions.
Core Features & Use Cases
- Bias auditing checklist: Walks a prioritized list of bias sources, from the AI editing skill template down to node descriptions, guidance fields, examples, and top-level constraints.
- Minimal-edit philosophy: Enforces the principle of aligning direction rather than prescribing rules, preferring removal of biased text over adding new guidance sections.
- Build pipeline integration: Covers regenerating the TOON schema with npm run generate:toon, validating with npm run check and npm run build, and testing via npm run debug.
- Use Case: When the AI editor keeps choosing sub-agent nodes for simple tasks, use this Skill to locate the biased description in the schema, make a minimal contrastive fix, regenerate the TOON file, and verify the corrected generation behavior.
Quick Start
Ask the AI to audit resources/workflow-schema.json for bias causing the wrong node type to be selected, then apply a minimal fix and regenerate the TOON schema.
