blogizi

Draft, update, and publish SEO blog posts to Blogizi from a local repo using the Blogizi CLI (preferred), hosted MCP as a sandbox backup, or the public API. Use when the user mentions Blogizi, blogizi draft/update/publish, MCP tools, posting markdown to a blog, SEO blog posts, or integrating with th

Blogizi@blogizi

Install

openclaw skills install @blogizi/blogizi

Blogizi CLI — AI Agent Guide

Help the human ship blog posts to Blogizi from their local repository. You (the coding agent) write the markdown in plain text; the CLI only authenticates and drafts/updates/publishes. Full human docs: README.md. Site overview: https://blogizi.com/llms.txt

Transport preference: CLI first → hosted MCP if the CLI cannot run (sandbox / no shell / no Node) → raw HTTP API only if neither is available.

Hard rules

Never ask the user to paste an API key into chat. Tell them to run blogizi auth <key> locally (or edit ~/.blogizi/config.json themselves). For MCP, tell them to put the key in their MCP client config — never in the chat.
Never commit ~/.blogizi/config.json, API keys, or .env secrets.
Default to drafts. Use blogizi draft (or MCP draft_post) for the first push unless the user explicitly asks to go live (blogizi publish / MCP publish_post).
Use blogizi update (or MCP update_post) to re-push edits to an existing slug. Do not invent a new slug just to avoid conflicts.
Do not invent Blogizi dashboard UI steps beyond Account → API and project Settings. Prefer CLI + README (or MCP docs when using tools).
Do not shell out to claude / codex / gemini via Blogizi. Write the content yourself (or with the user), then draft/update via CLI or MCP.

Prerequisites checklist

Before drafting or publishing, verify (run commands; don't assume):

node -v                    # need >= 20
which blogizi || npm install -g blogizi
blogizi --help
test -f ~/.blogizi/config.json || echo "NOT AUTHENTICATED"

If not authenticated:

text

Ask the user to:
1. Open https://blogizi.com → Dashboard → Account → API
2. Copy the account API key
3. Run locally: blogizi auth <api-key>
4. If they have multiple projects: blogizi use <project-slug>

If blogizi cannot be installed or run (sandbox, locked-down shell, no Node): skip the CLI checklist and use hosted MCP instead (see below). Do not fall back to inventing dashboard-only steps.

Core workflows

A. Write → draft → edit → update → publish (recommended)

# 1. You write ./posts/<slug>.md with Blogizi frontmatter (grounded in the repo)

# 2. Upload as draft
blogizi draft ./posts/<slug>.md

# 3. Improve the file with the user, then update the same slug
blogizi update ./posts/<slug>.md

# 4. Only when the user confirms "publish" / "make it live"
blogizi publish ./posts/<slug>.md

B. Multi-project accounts

blogizi use the-correct-project-slug
# or
blogizi auth <key> --project the-correct-project-slug

Wrong project is a common failure mode — confirm slug when the user has more than one blog.

Frontmatter pattern (required for draft/update/publish)

yaml

---
title: "How we built our auth flow"
description: "≤160 char meta description; specific, not clickbait"
keyword: "indie app authentication"
slug: how-we-built-our-auth-flow
status: "draft"
date: "2026-08-07"
readingTime: 0
wordCount: 0
---

Body markdown here. Do not repeat the title as an H1 — the blog UI already shows the title.

Writing tips

Ground the post in this repo (real APIs, files, decisions). Avoid generic filler.
slug: lowercase kebab-case; stable (changing slug creates a new post unless using blogizi update / API upsert on the old slug first).
description: one sentence, SEO-useful.
keyword: phrase the post should rank for.
status in the file is overridden by draft (→ draft) and publish (→ published). update only changes status when the file sets status explicitly; otherwise the existing post keeps its visibility.

When the CLI is unavailable (sandbox / no shell)

Prefer the hosted MCP over raw HTTP. Use this path when:

You cannot install or run blogizi (restricted sandbox, no network install, no Node)
The client has MCP tools but no reliable shell
The user already connected Blogizi MCP in Cursor / Claude / etc.

Docs: https://blogizi.com/docs/mcp

MCP connect (user configures once)

Streamable HTTP: https://blogizi.com/api/mcp with Authorization: Bearer <account API key>.

json

{
  "mcpServers": {
    "blogizi": {
      "url": "https://blogizi.com/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_ACCOUNT_API_KEY"
      }
    }
  }
}

Stdio-only clients can bridge with mcp-remote (see MCP docs). Same auth rules: user adds the key in client config — not in chat.

MCP tools (map 1:1 to CLI)

Goal	CLI	MCP tool
List blogs	(config / `blogizi use`)	`list_projects`
First push / new slug	`blogizi draft`	`draft_post`
Re-push same slug	`blogizi update`	`update_post`
Go live	`blogizi publish`	`publish_post`

Post tools take title, slug, optional description / keyword / content, and projectSlug when the account has more than one project. update_post accepts optional status; omit it to keep draft/published. draft_post fails if the slug exists — use update_post.

With MCP you still write the post body yourself; pass it as content (markdown without frontmatter). You do not need a local .md file if the sandbox has nowhere useful to write one — but prefer a file when the repo is writable so the user can review.

Raw Public API (last resort)

Only if neither CLI nor MCP tools are available (or the user wants integration code). Base: https://blogizi.com

http

Authorization: Bearer YOUR_ACCOUNT_API_KEY
X-Blogizi-Project: project-slug
Content-Type: application/json