Menu
Le prompt qui vérifie d'abord la source est sélectionné par défaut. Vous pouvez passer à une commande directe ou télécharger une copie locale.
Lisez SKILL.md et les fichiers associés affichés par SkillsMP avant de décider de l'installer.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.
npx skills add https://github.com/WeZZard/MyWorks --skill outlineLa commande reste sur une seule ligne. Faites défiler horizontalement pour la vérifier avant de la copier.
Vous préférez une copie locale ? Téléchargez les fichiers actuellement disponibles dans SkillsMP.
wpw entry point for cover images. Generate cover image(s) for a post on one or more surfaces (X Article, 微信公众号头条/图片消息, 小红书, generic ratios), then run preflight. Delegates generation to the `cover-image` skill from the pi-cover-image package; this shim only maps the post directory to the plugin's inputs, loops over surfaces, and chains preflight.
Run pre-publish checks on a blog post — frontmatter, content, feed-page exposure, and cover image checks per platform. Use before publishing a post under YYYY-MM-DD-slug/.
Polish English post prose by delegating to the `polish-editor` subagent. Use for blog post files under `YYYY-MM-DD-slug/index.md` or pasted English post text.
Basé sur la classification professionnelle SOC
Affichage de SKILL.md
| name | outline |
| description | Make an outline for a post. |
Use this skill as the shell around the outline-generator subagent.
This skill does not own outline design principles. Keep outline methodology in the outline-generator subagent. The skill only prepares context, invokes the subagent, and interprets the response for the user.
outline-generator.outline-generator subagent with the subagent tool and that prompt.MUST NOT:
Before invoking outline-generator, create a compact summary from the current conversation.
Include only information the subagent needs:
Call the subagent tool like this:
{
"agent": "outline-generator",
"task": "<spawning prompt>"
}
Pass the summary to outline-generator as an append-only input bundle.
Use this prompt shape:
<SPAWNING_PROMPT_TEMPLATE>
## Core thesis
<!-- What is the article trying to say in one sentence? -->
## Personal Stake
<!-- Why does this matter to the author specifically? -->
## Reader Promise
<!-- What will the reader understand better after reading this? -->
## Key Raw Materials
<!-- Important experiences, arguments, examples, quotes, contrasts, or observations. -->
## Tension / Conflict
<!-- What is the central disagreement, paradox, or unresolved problem? -->
## Constraints
<!-- Tone, register, audience, things to avoid, authenticity rules. -->
## Existing Post
<!-- The path to the existing post. Optional. Empty when the post is not created. -->
## Revisions
<!-- The historical revisions of the discussed outlines. You **MUST** list in the following format:
### 1. [Design purpose of the outline V1]
**Section Titles:**
1. <Section Title 1. Verbatim copy from the `outline-generator` subagent outputs.>
2. <Section Title 2. Verbatim copy from the `outline-generator` subagent outputs.>
**Writing Approach:**
1. <Section Title 1>: <Writing approach of `` >
:
...
...
---
:
:
...
...
-->
</SPAWNING_PROMPT_TEMPLATE>
MUST:
You MUST explicitly state "The user has no opinions for this version." in User's Opinions section if the user found no opinions for the particular version of outline. You MUST exhaust the list of the user's opinions for each version of the outline.
MUST NOT:
You MUST NOT write prose. You MUST NOT invent examples. You MUST NOT smooth out the author's voice.
MUST:
You MUST honestly present what the outline-generator subagent responded.
You MUST unwrap the outer markdown codeblock of the response before outputing it.
You MUST include all the sections in ONE SINGLE markdown codeblock.
MUST NOT:
You MUST NOT include each section in SEPARATE markdown codeblocks.