Real scheduling
Exact times per channel, weekly queue slots, or “best time”. A dispatcher with retries and back-off publishes on time and tells you exactly what failed and why.
Open source · AGPL-3.0 · Self-hosted
Markette is a publishing system for teams: write once, tailor each network, schedule at the moment your audience is there, and see what worked. Run it yourself, keep your accounts and data, and bring the AI you choose.
22 destinations · organizations, workspaces and teams · API and MCP server · Android, iOS and Chrome share
Hosted social tools decide what you may publish, when, and with which AI, and they change those rules whenever it suits their business. Markette is the opposite: a tool you run, whose code anyone can read, audit, improve and fork. Your audience data stays in your database, your tokens are encrypted at rest, and the AI is any OpenAI-compatible model you point it to — or none at all.
Exact times per channel, weekly queue slots, or “best time”. A dispatcher with retries and back-off publishes on time and tells you exactly what failed and why.
A week-by-hour model per channel starts from each network's typical rhythm and shifts toward your own engagement as data comes in. AI can suggest times and periods, and explain them.
One draft, a version per channel: character counts the way each network counts them, previews, titles, flairs, alt text — and links moved to the first comment where they hurt reach.
Organizations hold shared channels; workspaces keep brands or clients apart. Teams get channel access, approval workflows and their own queue. SSO per organization.
Impressions, likes, comments, shares and clicks are collected from every network that reports them, per post and per channel, and they feed the best-time engine.
A documented REST API with personal tokens, an MCP server so assistants can draft and schedule, Android/iOS apps with “Share with Markette” from any app, and a Chrome extension for any page.
Missing one? Request a network — every provider is a single Python module.
Docker images for the server and the web app, PostgreSQL and Redis. One compose file, one .env.
# get the compose stack
git clone https://github.com/markette-org/markette-deploy.git
cd markette-deploy && cp .env.example .env
# start it
docker compose up -d
docker compose exec api python manage.py createsuperuser
docker compose exec api python manage.py \
make_superadmin [email protected]
# open http://localhost:8080