Postiz, an open-source social media scheduler, is trending on GitHub, gaining about 950 stars in a day and sitting at roughly 38.9k stars in total. Its GitHub repository bills it as "The ultimate agentic social media scheduling tool," and the pitch that explains the surge is simple: it ships an MCP server, a command-line interface and an API so a coding agent can schedule and manage posts for you.
This guide covers what Postiz does, what it costs to run, what the self-hosting effort really involves, and where it fits next to the agent tooling we have covered.
TL;DR: Postiz at a glance
| Question | Answer |
|---|---|
| What is it? | Open-source social scheduler with analytics, team features and AI tools |
| License | AGPL-3.0 |
| Platforms | Instagram, YouTube, Dribbble, LinkedIn, Reddit, TikTok, Facebook, Pinterest, Threads, X, Slack, Discord, Mastodon, Bluesky (repo list) |
| Agent support | MCP server, CLI, public API, webhooks, Node.js SDK |
| Is it free? | Self-hosting is free; Postiz Cloud is a paid subscription with a 7-day trial |
| Hardest step | Creating and getting approved developer apps on each platform |
| Stack | pnpm workspaces, Next.js, NestJS, Prisma with PostgreSQL, Temporal, Resend |
What does Postiz actually do?
According to the repository, Postiz covers post scheduling, analytics and team collaboration, where members can comment on and schedule posts. It adds AI features: an AI Copilot, AI image and video generation, AI video clipping and a "Smart Agent." The documentation at docs.postiz.com describes publishing to 34 social platforms from one calendar, with per-platform tailoring of each post.
For automation, it offers a public API and webhooks, with integrations for n8n, Make.com and Zapier. The agent pieces are the newer draw: an MCP server and a CLI, with connectors for Claude, ChatGPT, Codex, OpenClaw and Cursor, plus a Node.js SDK on npm.
Why are developers excited about the MCP server?
An MCP server lets an assistant call tools in a standard way. With Postiz, the assistant can draft, schedule and manage posts as tool calls instead of you pasting text into a web dashboard. That fits a pattern we have followed across the site: coding agents moving beyond code into operations. Related examples include MCP events for event-driven agents, Talorys, a personal agent that runs on Cloudflare's free tier and cross-harness agent memory with Beacon.
On Cloud, there is a hosted MCP endpoint. Self-hosters point the MCP server or CLI at their own instance. We did not run the server for this post, so we cannot report latency, tool names or reliability. Check the docs' CLI and MCP pages before relying on a particular tool call.
How much does it cost?
There are two routes.
Self-hosted. Free under AGPL-3.0. You pay for your infrastructure, and for AI features you bring your own provider keys, such as an OpenAI key, and pay that provider directly. The repository says you can deploy with Docker, Coolify, Railway or any VPS, and that you are responsible for configuring Postgres, Redis, storage and environment variables. Docker Compose files and a Quick Start Guide are linked from the docs.
Postiz Cloud. A managed subscription with plan-based limits and a 7-day free trial. Postiz handles updates, backups and security. Channels, posts, team members and AI usage are capped by plan tier, and the AI features come with monthly quotas. We did not capture current prices, so check the pricing page.
What is the catch with self-hosting?
A post card with arrows fanning out, showing how a self-hosted Postiz scheduler publishes one post to many platforms
The repository is candid about it, and this is where most evaluations will stall.
- Platform developer apps. To post to each network you must create your own developer app and get it approved. The repository says Meta, YouTube and TikTok approvals can take weeks.
- Analytics credentials. Analytics on a self-hosted instance need your own app credentials with analytics scopes.
- Operations. You run the database, cache, storage and background workflow engine yourself, and you keep them updated.
- AGPL. The license is strong copyleft for network services. If you modify Postiz and offer it to others over a network, you generally must share your changes under the same license. Read the license with your lawyer if you plan to build a product on top of it.
What are the open questions?
- Reliability of agent posting. Automated posting is exactly where mistakes become public. Run agents in draft mode first and require approval before publication.
- Platform terms. Social networks restrict automated and AI-generated posting in different ways. A scheduler does not exempt you from each platform's rules, and some networks limit API access to approved apps.
- Maintenance load. The page shows 135 open issues and 87 open pull requests, in line with a busy project, but it does not show a latest release, so check release history before pinning a version.
- Security of agent access. Giving an agent credentials to post under your brand is a risk. Scope keys narrowly and keep them out of prompts. For broader agent risk, see AgentBeam, the agent security platform from the explainx.ai team that stops AI agents before they take dangerous actions.
What are people asking about Postiz?
How is it different from a hosted scheduler? Hosted schedulers handle the platform developer apps for you, because they already hold approved integrations. With self-hosted Postiz, that approval burden moves to you, which is the trade for owning the data, the code and the cost curve. Postiz Cloud exists for people who want the hosted convenience with the open-source codebase behind it.
Do I need an AI provider key? Only for the AI features when self-hosting. The repository says self-hosted users supply their own provider API keys and pay the provider directly, while Cloud bundles the AI features with monthly quotas. If you only schedule posts you write yourself, you can ignore the AI parts.
Can I run it without an agent? Yes. The calendar, analytics and team features work as a normal web app, and the agent interfaces are optional layers on top.
What does an agent workflow look like? A plausible, low-risk pattern, which is our suggestion and not an official recipe, is to have the agent turn a release note or blog post into three platform-specific drafts, place them in the calendar as drafts, and stop. A person then edits and approves. A prompt along the lines of "draft a LinkedIn post and an X post announcing this release, using the post's tone, and save both as drafts for tomorrow morning" keeps the human in charge of what goes public.
Is AGPL a problem for personal use? For running Postiz for yourself or your own company, the license obligations are light. They matter most if you modify the code and let others use it over a network, or if you want to embed it in a commercial service. When in doubt, read the license text on the repository.
How does it fit with the rest of the agent stack?
Postiz is a good example of a category that is growing: ordinary business software that exposes itself to agents through MCP and a CLI instead of only through a dashboard. The practical advantage is that your coding agent can reuse the same interface for many tools. The practical risk is that every added tool is another set of credentials an agent can misuse. Treat each connector the way you would treat a new employee's access: scoped, logged and reversible.
If you already use a coding agent for release work, the natural place for Postiz is the last step of the pipeline, after tests pass and release notes are written. Keep publishing behind an approval gate so a hallucinated claim never goes out under your name, and keep a record of what the agent queued and why.
Who should use Postiz?
| You are | Fit |
|---|---|
| A solo creator or small team who wants an open scheduler and is comfortable with Docker | Good, if you can wait through platform approvals |
| A team that wants an agent to draft and queue posts | Good, with a human approval step |
| A business that wants zero maintenance | Consider Postiz Cloud rather than self-hosting |
| Someone who only needs a quick post | A native scheduler on each network is faster |
A sensible first test
- Start with Postiz Cloud's trial, or a throwaway self-hosted instance, and connect one low-risk account.
- Connect an agent through the MCP server or CLI and ask it to create a draft, not a published post.
- Review what the agent produced and whether the post text, media and schedule match your intent.
- Only then connect brand accounts, and keep a human in the approval loop.
What we could not verify
We read the repository page and the documentation introduction. We did not install Postiz, run the MCP server or confirm current pricing, and the repository page did not show a latest release. Star counts and platform lists are as shown at the time of writing and change often.
Related reading
- MCP events and event-driven agents
- Talorys personal AI agent on Cloudflare's free tier
- Beacon: cross-harness agent memory
- Scientific Agent Skills: 177 research skills
- Aurelio Finance: open-source self-hosted AI advisor
- XERJ autoindex: local search for agents
Details reflect the Postiz repository and documentation on October 27, 2026 and may change.