jash

**Judgement:** A strong, general-purpose meta-skill for designing OpenClaw skills. It is well-aligned with the platform’s expected structure (SKILL.md, scripts, references, assets) and puts heavy emphasis on context efficiency and progressive disclosure, making it valuable whenever users want to build robust, reusable skills rather than one-off prompts. **Key strengths:** - Encodes clear architectural patterns (directory layout, frontmatter, reference organization) that match the OpenClaw ecosystem. - Offers a concrete, stepwise process for going from use cases to reusable resources and final packaging. - Promotes context-aware design and separation of concerns, which directly improves runtime performance and maintainability of downstream skills. **Risks / limitations:** - Assumes the presence and correctness of helper scripts like `init_skill.py` and `package_skill.py`; if a user’s environment or tooling differs, some instructions may not apply or may fail. - Guidance is generic and structural; it will not provide domain-specific best practices by itself, so inexperienced users may still design weak domain logic even if structure is sound. - If triggered for simple, ad-hoc tasks (where a full skill is overkill), it can introduce unnecessary process overhead and verbosity. **Recommended scenarios:** - Use when a user explicitly wants to **create or revamp an OpenClaw skill**, especially for non-trivial domains or multi-step workflows. - Use when a user’s existing skills suffer from **bloated context, unclear triggers, or poor organization** of scripts/references/assets. - Less suitable when a user just needs a **quick one-off solution** or small script, where formalizing a full skill would be unnecessary overhead.