Source discussion
Problem
Users are trying to adapt ECC to non-native harnesses like Grok/xAI by manually copying agents, skills, and rules into prompts or custom instructions. We do not yet have a canonical manual-adaptation guide for harnesses that do not support the native ECC folder layouts.
Proposed outcome
Document and/or package a fallback workflow for non-native harnesses that covers:
- how to pick the minimal agent/skill/rule set for a task
- how to compress ECC context into a manual master prompt without dragging the whole repo in
- how to reproduce the value of commands and hooks when the target harness has no native support
- what the limits are compared with first-class targets like Claude, Cursor, Codex, Opencode, and CodeBuddy
Definition of done
- there is a documented ECC-native adaptation path for manual / non-native harnesses
- the answer is reusable for Grok and similar tools instead of living in one-off discussion replies
Source discussion
Problem
Users are trying to adapt ECC to non-native harnesses like Grok/xAI by manually copying agents, skills, and rules into prompts or custom instructions. We do not yet have a canonical manual-adaptation guide for harnesses that do not support the native ECC folder layouts.
Proposed outcome
Document and/or package a fallback workflow for non-native harnesses that covers:
Definition of done