blogwriter-mcp

Write blog posts from your material and in your voice. Your agent gathers the material (your thesis, your experiences, claims it verified), this server sends it to a separate model that writes the post, and the agent checks the result before sending corrections back.

Install with Claude Code

/plugin marketplace add paraspace-dev/writer-mcps
/plugin install blogwriter@writer-mcps

Claude Code prompts once for an OpenAI API key and stores it in the OS keychain. The plugin registers the blogwriter MCP server, a skill that keeps the agent in the material-in role, and a command to set up project writing configuration:

/blogwriter:blogwriter-init

How it works

The agent calls create for a new post, then checks the result against its material. If it finds an error, it calls edit with the corrected material rather than rewriting the prose itself.

Use review when you want editorial findings instead of a rewrite. It returns findings with locations, or reports that the post has no significant editorial problems.

The check matters most for first person. The writing model may only recount experiences you actually supplied; the agent verifies that nothing in your voice was invented, the way it fact-checks commands and links.

blogwriter-mcp writes tutorials, explainers, opinion pieces, stories, announcements, and retrospectives. The first line of every post is its title as a # heading; keep it, or move it into your blog's front matter.

Other MCP hosts

The server is the npm package blogwriter-mcp. Run it as npx -y blogwriter-mcp@latest with OPENAI_API_KEY in its environment.

Claude Code project configuration

For Claude Code without the plugin, add the server to .mcp.json:

{
  "mcpServers": {
    "blogwriter": {
      "command": "npx",
      "args": ["-y", "blogwriter-mcp@latest"]
    }
  }
}

Codex

For Codex, add the server to ~/.codex/config.toml:

[mcp_servers.blogwriter]
command = "npx"
args = ["-y", "blogwriter-mcp@latest"]

Give it your voice

Configuration is optional. Without .blogwriter-mcp/, the server uses its editorial brief and a default model.

Add examples and writing rules when you want posts that sound like you:

.blogwriter-mcp/
  config.toml       model choice, voice corpus globs
  instructions.md   your house style, in your words
  voice/
    examples/       posts that sound the way you want
    avoid/          counterexamples of the style you keep deleting

[model]
model = "gpt-5.6-terra"

[voice]
include = ["posts/**/*.md"]
exclude = ["posts/drafts/**"]

Files in voice/ and files matched by include become the voice corpus. A few strong examples usually teach the model more than a long instructions file. Pre-AI posts make the best examples; if your published posts are AI-written, adding them to include teaches the model that same voice. AI originals recovered from git history can go in voice/avoid/.

The corpus is capped at max_kb, which defaults to 256 KB. The server reports files it drops for the budget. The corpus is sent with every request and uses the provider's prompt cache, so it is nearly free after the first call of a session. Set BLOGWRITER_MCP_DEBUG=1 to print input, cached, and output token counts to stderr.