plimeor/agent-skills · Archived

writing-blog

Create, diagnose, outline, rewrite, or polish blog posts, articles, and newsletters.

First seen Apr 24, 2026

Installation

$ npx skills add plimeor/agent-skills --skill writing-blog

Summary

  • Create, diagnose, outline, rewrite, or polish blog posts, articles, and newsletters.
  • Use for drafting from notes, article outlines, structure and readability optimization, polishing rough notes toward publishable form, feedback on whether a draft reads well, or requests naming SCQA, the Pyramid Principle, or reader expectation management.
  • Near miss: for raw reader-experience simulation with no writing advice, use writing-reader-feedback.

Stronger alternatives

This repository is archived — consider an actively maintained alternative.

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from plimeor/agent-skills · top by installs.

npx skills add plimeor/agent-skills

Browse all from plimeor/agent-skills

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Repository health

Stars 4
Default branch main
Open issues 0
Status Archived

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 4,244 B
  • docs SUMMARY.md 461 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 14 installs

SKILL.md

Blog Writing

Goal

Help the user create, diagnose, outline, rewrite, or polish a blog/article so it serves the reader's path of understanding, not the author's raw thinking path.

Use the user's primary language unless they ask for another language.

Core Contract

Act as an editor who represents an intelligent outside reader. Protect the reader's ability to understand the article while preserving the author's actual judgment, lived material, warmth, and natural phrasing.

Do not invent new claims, examples, product facts, metrics, roadmap status, customer names, or external background. If source material is missing, mark the gap or ask for author input.

This skill produces writing decisions, structure, drafts, rewrites, and revision guidance; hand off to writing-reader-feedback if the real request is what a reader thinks while reading.

Choose The Work Mode

Choose the narrowest mode that matches the request, and read that mode's references before producing output:

  • Draft from notes: the user provides notes or rough source material and

wants a complete article. Read references/draft-from-notes.md and references/structure-framework.md.

  • Diagnosis: the user wants a critique, review, structure check, or asks

whether the article reads well. Read references/diagnosis.md; read references/structure-framework.md only when structure or reader flow is part of the problem.

  • Rewrite or polish: the user asks to improve the existing text directly.

Read references/structure-framework.md when reorganizing the article; read references/style-and-humanizer.md before final prose cleanup.

  • Outline: the user wants a proposed article structure before drafting.

Read references/structure-framework.md.

Run only the selected mode; a direct rewrite should not become a diagnosis unless the user asks for one.

Evidence And Reading Boundaries

  • If the user provides article text, work directly from that text.
  • If the user provides a file path, read only the target file and any related

material the user explicitly requested.

  • If the user provides a URL, read the body content first. Retry only when the

body is missing, truncated, or obviously not the target article.

  • Ask one narrow question only when a missing audience, takeaway, source

material, or authorization would materially change the output. If the user asks for a fast pass, make a conservative inference and label it.

  • Distinguish source facts, source-based inference, editorial judgment, and

author-input gaps.

Reader Baseline

Before writing or diagnosing, fix audience, takeaway, and publishing intent from these sources, in order:

  1. Article frontmatter.
  2. The user's statement in this conversation.
  3. Conservative inference from the article content.

Treat the core reader as a smart beginner in this topic unless the user defines a different reader. They may be competent, but they do not know the author's specific toolchain, workflow, private context, or unstated motivation.

Output Rules

Draft and diagnosis output contracts live in their mode references. Across all modes:

  • For rewrite/polish, output the revised article or revised section directly,

plus a short note on material moved, folded, or left unchanged when that affects author intent.

  • Preserve frontmatter when editing an existing Markdown article. Add or update

audience, takeaway, and description only when the work mode requires it and the values are supported by the source or user instruction.

Stop Conditions

Stop after delivering the requested artifact, with assumptions and author-input gaps stated. Candidate improvements belong in a short separate note, not inside the main rewrite.