openclaw/human-writing

Human-Writing Skill

This skill provides the operational standards for generating professional, high-density, and human-sounding documentation and prose. It is designed to purge common LLM stylistic "tells" and replace them with the precision of a senior engineer or domain expert.

Core Directives

1. Eliminate AI "Tells": Before finalizing any documentation or formal text, refer to ai-tells.md to identify and remove overused LLM vocabulary, structural tropes, and puffy language.
2. Apply Professional Standards: Follow the guidelines in standards.md for technical precision, information density, and tone.
3. No Buzzwords: Zero tolerance for "synergy," "cutting-edge," "revolutionize," "seamless," or "leverage." If a technical term exists, use it.
4. No Emojis in Docs: Reserve emojis for chat interactions (as per SOUL.md). Professional documentation (READMEs, PR descriptions, code comments) must remain text-only for maximum clarity.
5. Precision Over Prose: Humans in the field value numbers, versions, and RFCs over flowery descriptions.

Workflow

When asked to "write documentation," "create a README," or "explain this technically":

1. Scan references/ai-tells.md for words to ban from the current draft.
2. Apply references/standards.md to structure the output with high density and low fluff.
3. Draft the content.
4. Self-Audit: Verify the output does not contain "Rule of Three" adjectives or "Not only... but also" parallelisms.
5. Finalize: Remove all emojis and corporate filler.

Reference Materials

• ai-tells.md - Field guide to AI writing "tells" to avoid.
• standards.md - Human-like professional writing standards.