prompt-and-skill-lifecycle

This reusable Skill connects instruction design with ownership, delivery, and evidence from use. It helps author, audit, organize, publish, and maintain Prompts and Skill packages while keeping their meaning and actual consumers visible together.

Its method connects purpose → relational form → working realization: state what the actual consumer should accomplish, explain the relevant relations and conditions, and give them a usable form in instructions and interfaces. These are connected perspectives, not mandatory stages. Read observed behavior back through them when deciding what to revise. Model-facing instructions normally state purpose, roles, actions, and judgment criteria directly. Philosophical or literary expression can clarify meaning that direct instruction has not conveyed; corresponding expressions can inform and revise one another through use.

Adopt it

Keep this complete directory together. Use your host's documented Skill installation or discovery mechanism to expose SKILL.md; verify that the consumer can retrieve it and the referenced resources. A checkout or ZIP alone is not an installation. Hosts without native Skills can read the entrypoint and selected resources as task guidance, within their own instruction hierarchy. This package contains no installer, service, or required tool API.

Typical requests include “audit this project's instructions for conflicting ownership,” “rewrite this Skill for a different consumer,” or “check whether the published package is actually available and used.” Begin at the entrypoint, which routes to semantic authoring, integrated change, delivery, or usage evidence. Read only what the current action needs.

Scope and adaptation

The method works without the original author's computer, repositories, private documents, model selections, or publication services. Adapt discovery metadata and delivery to your real host. Preserve its existing authority, authentication, consent, and stopping boundaries; this Skill adds no mandatory approval or Plan process. Yinyang is an optional related example, not a dependency.

The public edition retains the substantive semantic method, single ownership, integrated revision, proportionate evidence, and consumer readback. It replaces the original host profiles and synchronization scripts with portable questions about actual ownership and delivery. Private participant models and fixed conversation infrastructure are excluded. The two external authoring articles are attributed in Authoring sources, rather than copied into the package; current model facts need their own verification.

Package/link checks establish source readiness. Installation and effectiveness remain claims to verify in your actual consumer. Distinguish discovery, retrieval, application, and outcome rather than treating any one as the others.

This independent package uses the MIT License. You may obtain the complete directory from GitHub and use, modify, and redistribute it, including commercially, with the required MIT notices. It does not include Yinyang. The public collection provides downloads and Issue discussion; this collection does not accept pull requests.