How to use the Blog Draft Writer agent skill (blog-draft)

What the blog-draft skill makes the agent do

The skill walks the agent through a fixed sequence of steps. The agent first gathers resources such as URLs, files or topics (using web search for topics and URLs), and saves a summary of each one. It then brainstorms angles and asks you clarifying questions, proposes an outline, and waits for your approval before writing.

After the outline is approved, the agent writes a full draft in which every statistic, comparison and factual claim carries an inline citation, with a References list at the end. Each draft is saved as a new numbered version, such as draft-v0.1.md, draft-v0.2.md and so on, so you can compare revisions or go back to an earlier one.

• Resource summaries are saved in a resources/ subfolder.
• The approved outline is saved as OUTLINE.md.
• Drafts are saved as draft-v0.1.md, draft-v0.2.md, and so on.
• The agent pauses for your feedback at each checkpoint and does not skip them.

Install the blog-draft skill

Download the blog-draft ZIP from the skill's page on this directory. You can install it in one of two ways, depending on how you use your agent.

For Claude Code, unzip the download into your personal skills folder ~/.claude/skills/ to make it available everywhere, or into a project's .claude/skills/ folder to limit it to that project.

In the Claude apps, upload the skill under Settings → Capabilities → Skills instead.

• Personal install: unzip into ~/.claude/skills/
• Project install: unzip into the project's .claude/skills/ folder
• Claude apps: Settings → Capabilities → Skills, then upload the skill

Check what you need before you start

The skill needs an AI agent that supports SKILL.md skills. It also needs web search access, because the instructions tell the agent to use it for topics and URLs. Git and a remote repository are only needed if you want the commit and push steps.

The skill does not run scripts. It does use the network: it fetches URLs you give it, runs web searches, and pushes commits to a remote when you are in a git repository.

• An agent that supports SKILL.md skills
• Web search access for topics and URLs
• Git and a remote repository, only if you want commits and pushes
• Runs scripts: no. Uses network: yes.

How to trigger the skill with natural requests

You do not need a special command. Describe what you want in plain language and include the topic, and ideally your resources, target audience, and tone. The skill's description says it is used when writing blog posts, creating content from research, and drafting articles.

If you already have a blog post and want to revise it, say so. The skill then skips the early steps, reads your existing draft file first, and starts at the revision step.

• "Write a blog post about remote onboarding. Use these three links as sources. Audience: engineering managers. Tone: plain and practical."
• "Draft a blog post on how solar batteries work for homeowners. Keep it around 1000 words."
• "I have an existing blog post in my blog-posts folder. Revise it based on this feedback: shorten the intro."

Work through the checkpoints

The workflow creates a project folder named with the date and a short topic slug, such as YYYY-MM-DD-short-topic-name, and asks you to confirm before creating it. Sources are summarised first, then you see the outline. The skill asks you to approve or correct the outline before it writes the draft.

After each draft, the agent asks for feedback on overall impression, sections to expand or cut, tone, missing information and specific edits. Ask for changes and the agent saves the next version. When you approve, it can rename the final version to final.md if you request it, and summarise the number of versions, the main changes, the final word count and the files created.

• Target length options: short (500-800 words), medium (1000-1500), long (2000+)
• Folder layout: blog-posts/ with resources/, OUTLINE.md and draft-v0.1.md, draft-v0.2.md and so on
• Git commits use messages like docs: Add outline for blog post - [topic-name] and docs: Add draft v0.1 for blog post - [topic-name]

Check citations and safety before you publish

The skill asks for citations on every data point, but the agent can still make errors or misattribute a source. Open each cited source and confirm the claim before publishing.

If you are working inside a git repository, the skill commits and pushes to the remote. If that remote is connected to a publishing system, files may be published unexpectedly. Use a repository you are comfortable pushing to, or work outside a git repository.

• Verify every statistic and quote against its original source.
• Check what the remote does on push before using the git steps.
• Note that the SKILL.md instructions are written in Japanese, so check that the agent output matches the language you need.