prompt-engineering

To convert a recurring workflow into a reusable SKILL.md file, this Skill uses a structured four-round interview to extract goals, step breakdowns, arguments, and edge cases, generating a complete definition with frontmatter, when_to_use triggers, and per-step success criteria.

Authoring agent personas and system prompts involves creating structured SKILL.md templates with clear frontmatter rules, explicit argument hints, and annotated success criteria to ensure safe, consistent, and repeatable automation behavior across developer and operations teams.

Yes, you can enforce specific allowed-tools when generating skill definitions. The authoring process respects project override files like user-config.md to strictly control which tools the automation can access during execution.

The interview process for capturing ad-hoc automation logic works through a four-round pattern that progressively extracts workflow goals, breaks down steps, defines arguments, and identifies edge cases to produce a complete and discoverable SKILL.md definition.

You need a standard project directory structure to save generated skill configurations. The process outputs completed definitions directly into a .claude/skills/<name>/SKILL.md file path for immediate discoverability and operational use.

Limitations when converting ad-hoc workflows into repeatable skills include the strict dependency on a structured interview pattern for accurate extraction and the need for existing project override files to properly enforce allowed-tools and when_to_use triggers.