skill-prompterlisted

Prompting compliance specialist for AI agent skills — rewrites SKILL.md steps with RFC 2119 wording, explicit tools, VERIFY gates, and dry-run checks; also owns skill name + frontmatter description wording (trigger-surface rules). Use only when the user asks to improve or harden a skill's prompts/steps or to name/review a skill description, and confirm with the user before running since this takes quite long time and need to run a few smoke tests with subagents to check if the skill is perfect. Structural lifecycle work goes to skill-manager.
kargnas/skills · ★ 6 · AI & Automation · score 66

Install: claude install-skill kargnas/skills

# Skill Prompter > **ONE SENTENCE:** Rewrite any AI agent skill's step wording so every step is executed 100% — even by weaker LLMs that skip steps. --- When starting, let the user know in one line that this skill is being used. --- ## Scope and Boundaries Two skills cover the skill lifecycle. Use the right one — they are not interchangeable. | Concern | Owner | |---------|-------| | Step wording, RFC 2119 keywords, atomic decomposition, tool naming, VERIFY gates, decision tables, subagent delegation, escape hatches, dry-run compliance, skill `name` + `description` wording (trigger surface) | **skill-prompter (this skill)** | | Skill creation lifecycle, directory layout, file separation, references/scripts/assets organization, modular architecture, merge/split decisions, anti-patterns of structure | **skill-manager** | **Decision rule:** - "Steps are being skipped on weaker models" → skill-prompter - "I need a new skill from scratch" → skill-manager (then skill-prompter to harden the steps) - "Two skills should be merged" or "this skill is too big" → skill-manager - "Step instructions are vague" → skill-prompter - "Name this skill" or "write/review this skill's description" → skill-prompter (read [references/description-rules.md](references/description-rules.md)) - Both axes need work → run skill-manager first (structure), then skill-prompter (wording), then dry-run If the request is purely structural, **delegate to skill-manager** instead of proceeding here. --- #