$ man content-wiki/content-mcps

Model Context Protocol servers give your AI agent the ability to interact with external platforms. For content creation, this means the agent in your code editor can directly publish to Typefully, create drafts on Substack, read and post to Slack, interact with LinkedIn through browser automation, and manage your entire content pipeline without you ever leaving the IDE. The shift is fundamental: instead of writing content in your editor, then copying it to each platform manually, the content OS handles distribution as part of the pipeline. Write once, publish everywhere, all orchestrated by agent skills that call MCP tools.

PATTERN

Typefully MCP: drafting and scheduling posts for X and LinkedIn. Create drafts, schedule publication times, manage the content queue. This is the primary publishing pipeline for social content. Substack MCP: creating newsletter drafts directly on Substack. The final-substack skill uses this to push finalized posts without opening the Substack editor. Handles title, subtitle, and body content. Slack MCP: reading and posting to Slack channels. Used for partner communications, syncing channel history to markdown, posting reminders, and sending content updates to team channels. Browser-use MCP: browser automation for platforms without APIs. Used for LinkedIn interactions — reading posts, posting comments, checking engagement metrics. The browser agent navigates LinkedIn like a human, which is the only reliable way to interact with LinkedIn programmatically. Vercel MCP: deployment pipeline for the website. After making content or feature changes to the site, deploy directly from the editor.

PATTERN

Example workflow — publish a LinkedIn play: (1) Write the post in content/drafts/ as markdown. (2) Run final-copy to normalize voice and format. (3) Typefully MCP creates the draft on LinkedIn. (4) Browser-use MCP opens the published post to monitor engagement. (5) Slack MCP posts a notification to the content channel. (6) Daily tracker skill logs the publication. Example workflow — newsletter publish: (1) Write the post in content/substack/drafts/. (2) Run final-substack to normalize and push to Substack via MCP. (3) After publishing, the skill generates LinkedIn and X cross-promo snippets. (4) Typefully MCP schedules the cross-promo posts for the next day. Each workflow is a chain of MCP calls orchestrated by skills. The agent knows which MCPs to call, in what order, with what data. You trigger it with a single slash command.

PRO TIP

Any platform with an API can become an MCP server. The pattern: identify the API endpoints you need (create draft, schedule, publish). Write an MCP server that exposes those endpoints as tools. Configure it in your Cursor MCP settings. Now your agent can call those tools directly. For content platforms specifically: most expose create and read endpoints. That covers drafting and publishing. Some expose analytics endpoints — that lets your agent pull performance data and factor it into content decisions. The meta-pattern: every new platform you add to your content distribution expands the operating system's reach without changing any existing workflows. Add the MCP, update the relevant skill to include it in the pipeline, done. The architecture scales horizontally.