Publish your blog from any AI agent
- Guides for Claude, ChatGPT, Codex, Cursor, OpenClaw and more
- Three ways in: the MCP server, the skill + CLI, or the Claude connector
- Drafts on your own domain; you approve before anything goes live
AI agents can write, update and publish posts on your own domain. Blogizi supports them through a hosted MCP server, a CLI with an installable skill, and an official Claude connector. Keep a human in the loop: agents draft, you publish.
Which AI agents can publish to Blogizi?
Every agent below has a dedicated guide. Each row shows how it connects to Blogizi and how it signs in.
| Agent | Made by | How it connects | Sign-in | Guide |
|---|---|---|---|---|
| Anthropic | Official directory connector | Sign-in | Guide → | |
| Anthropic | Blogizi plugin, skill + CLI, or MCP | Sign-in or blogizi login | Guide → | |
| Anthropic | Directory connector, custom connector or plugin | Sign-in | Guide → | |
| OpenAI | Custom MCP connector (Developer mode) | Sign-in | Guide → | |
| OpenAI | Skill + CLI, or MCP server | blogizi login or codex mcp login | Guide → | |
| Anysphere | MCP in mcp.json, or skill + CLI | Sign-in | Guide → | |
| Nous Research | MCP in config.yaml, or skill + CLI | hermes mcp login | Guide → | |
| Open source | ClawHub skill + CLI (not MCP) | blogizi login or API key | Guide → | |
| Open source | MCP on one group, or skill + CLI | Key in the credential vault | Guide → | |
| Open source | Skill + CLI where agents run | blogizi login or API key | Guide → | |
| Perplexity | Custom connector | Sign-in | Guide → | |
| — | The MCP link | Sign-in or Bearer key | Guide → |
Not listed? Any MCP-compatible tool can use the MCP server, and any agent that runs shell commands can use the skill + CLI.
How do AI agents connect to Blogizi?
Three surfaces, one account. Whichever your agent supports, it reaches the same blogs and drafts.
| Blogizi MCP server | Blogizi skill + CLI | Claude connector + plugin | |
|---|---|---|---|
| Install | https://blogizi.com/api/mcp | npx skills add bytemindlab/blogizi-agent | claude.ai/directory/blogizi |
| Used by | Claude Code, ChatGPT, Cursor, Codex, Hermes, NanoClaw, Perplexity, any MCP client | OpenClaw, Paperclip, Codex, Claude Code and any agent with a shell | Claude, Claude Cowork, Claude Code |
| Sign-in | Sign in to Blogizi, or a Bearer key | blogizi login, or BLOGIZI_API_KEY | Sign in to Blogizi |
| Images and tools | All six tools, incl. Search Console (Pro) | Markdown files, relative images auto-uploaded | Same tools as MCP; the plugin adds the skill |
More on the Blogizi MCP server and the Claude connector.
One Blogizi account powers every blog your agents write for
Set up your blogs once. Then any agent, through any surface, can draft to them, on a subdomain or under yourapp.com/blog.
- list_projects
- list_posts
- draft_post
- update_post
- publish_post
- upload_post_image
- get_search_performance
How do I connect any AI agent to Blogizi?
Three steps, about two minutes.
Pick a surface your agent supports
MCP-native agents take the link. Agents with a shell take the skill and CLI. Claude chat users add the directory connector.
# MCP serverhttps://blogizi.com/api/mcp # or the skill + CLInpx skills add bytemindlab/blogizi-agentnpm install -g blogizi
Sign in to Blogizi
MCP clients open Blogizi sign-in. The CLI signs in with a device code. An API key is only for CI and headless servers.
blogizi login --no-browserblogizi status # check which blog is active # headless / CIexport BLOGIZI_API_KEY=your_api_key
Ask the agent to write
Describe the post in plain language. The agent drafts it on your domain; you review before publishing.
> Write a post about our new export feature, target "export notion to markdown", with the screenshot in ./docs/export.png as the cover. Agent: Drafted "Export Notion to Markdown in one click" (980 words) on blog.yourapp.com. Publish it, or want changes first?
How does the Blogizi CLI work for AI agents?
The CLI is the simplest way for an agent to publish: it writes a markdown file with frontmatter and runs one command. The skill teaches it when and how.
Pick the blog
Sign in once, then choose which blog the agent writes for. One account covers every blog.
blogizi login --no-browserblogizi statusblogizi use docs-blog # switch blogs
Write and draft
The agent writes markdown with frontmatter. One command saves it as a draft and uploads relative images.
---title: "Export Notion to Markdown in one click"description: "How our new export works"keyword: "export notion to markdown"slug: export-notion-to-markdown--- Your markdown here.  $ blogizi draft posts/export.md
Update, upload, publish
Re-push edits to the same URL, add images, and publish when it's approved.
blogizi update posts/export.md # same slugblogizi upload ./cover.png --slug export-notion-to-markdownblogizi publish posts/export.md # only when approved
Blogizi CLI command reference
Everything the CLI does. draft forces status draft, publish forces published, update keeps the current status unless the frontmatter sets it.
| Capability | Description |
|---|---|
| blogizi login [--no-browser] | Sign in with a device code; the session refreshes itself |
| blogizi logout | Sign out and remove the local session |
| blogizi status | Show the signed-in account and active blog |
| blogizi auth <apiKey> | Use an account API key instead of signing in (CI) |
| blogizi use <projectSlug> | Switch which blog commands write to |
| blogizi draft <file.md> | Save the markdown file as a draft; uploads relative images |
| blogizi update <file.md> | Re-push edits to the same slug |
| blogizi publish <file.md> | Publish the post |
| blogizi upload <file> --slug <post> | Upload an image to a post |
Run blogizi --help for the full list of flags.
What can an AI agent do with Blogizi?
Pair the CLI with the MCP server's get_search_performance and your agent can pick topics from real search data (Pro).
Write and draft
Agents save posts as drafts on your own domain.
Update by slug
Re-push edits to the same URL. No duplicates.
Images included
Covers and inline images upload with the post.
Publish on approval
Publishing is always a separate, explicit step.
Every blog you run
One account, many blogs; agents pick by name.
Search Console
Agents choose topics from real search data (Pro).
Can an AI agent write posts on a schedule without me?
Yes. Always-on agents run recurring jobs, and coding agents run headless in CI. Keep a human approving what goes live.
Always-on agents
Hermes, OpenClaw, NanoClaw, Paperclip
Give the agent a weekly job. It reads your Search Console data through Blogizi, drafts posts for page-2 queries and messages you the links (Pro).
You: Every Monday, draft posts for my two best page-2 queries and send me the links. Agent: Scheduled. This week: - "markdown blog for saas" → draft - "add blog to nextjs" → draft Reply "publish 1" or "publish 2".
Headless in CI
Codex and Claude Code
Run one prompt non-interactively after every release. The agent reads the changes and drafts the release post, with BLOGIZI_API_KEY set in CI.
# after a new tag is pushed$ codex exec "Summarize the changes since the last tag into a release post and draft it on Blogizi" # or$ claude -p "Draft the release post on Blogizi"
Why AI agents work well with Blogizi
- ✓Markdown in, a ranked post on your domain out
- ✓The skill teaches agents to draft first, never paste keys
- ✓Update by slug: no duplicate URLs when an agent re-pushes
- ✓Search Console data for picking topics (Pro)
- ✓SEO handled: sitemaps, schema, OG images, canonicals
- ✓Model-agnostic: works with whichever model your agent uses
list_projects every blog you rundraft_post save a draftupdate_post edit in place, same URLupload_post_image cover + inline imagespublish_post go live, when you say soget_search_performance Search Console (Pro)
What can I ask your agent to write?
Anything you'd brief a writer on. Name the topic, the keyword and the tone; your agent handles the rest through Blogizi.
Any agent
Draft
Same URL
Update
blogizi upload
Image
blogizi use
Switch blog
Where can your agent publish?
On your own domain, wherever your app is built. Point a subdomain or a /blog path at Blogizi once, then publish from your agent.
Next.js
Lovable
Vercel
Webflow
Framer
Replit
Bolt
v0
Astro
- subSubdomain
- /b/blog path
- ↪Any reverse proxy
Popular next steps
They're different products and Blogizi works with all of them. This page is for your agent.
your agent (this page)
The Blogizi CLI and skill, for every agent with a shell.
Claude Code
Anthropic's terminal coding agent. Blog from the repo you're already in.
Codex
OpenAI's coding agent for the terminal and IDE.
OpenClaw
Open-source personal agent with ClawHub skills.
Using another agent? See all agents.
Pricing, security, and troubleshooting
How much does it cost?
The MCP server, skill and CLI are included with every Blogizi plan. Your agent is billed by your AI provider. Blogizi starts with a 7-day trial for $0, then plans run from $9/month. The Search Console loop needs Pro or Agency.
Compare Blogizi plansHow is access secured?
blogizi login keeps a self-refreshing token in ~/.blogizi/config.json (mode 600). Run blogizi logout or disconnect it in Account → Connected apps.
Something not working?
- ✓Command not found: npm install -g blogizi (Node 20+).
- ✓Not authenticated: blogizi login.
- ✓Wrong blog: blogizi use <slug>.
Frequently asked questions
Your AI agent writes the post, and Blogizi drafts and publishes it on your own domain with SEO handled. You approve before anything goes live.
Give your agent a blog
Connect any AI agent to Blogizi and let it draft posts on your own domain. You approve; it does the rest.
Ready to get started?
Your agents write. Blogizi publishes. Google sends the traffic to your domain.