The developer had a simple problem. Every time he published a post on his EmDash CMS blog, he had to manually copy it to Medium, Dev.to, and LinkedIn. Three platforms, three browser tabs, three rounds of formatting fixes. A 1,500-word article took 45 minutes to syndicate. He built a solution using EmDash\u2019s built-in MCP server. It cut that time to under two minutes.<\/p>\n
Syndication is a standard workflow for content creators. Write once, republish everywhere. But the manual version is brittle. Copy-paste errors, broken image links, inconsistent formatting. The developer was spending roughly three hours a week on a task that added no editorial value. He wanted a system that would, on publish, automatically push the content to each platform with correct metadata.<\/p>\n
He had tried RSS-based automation tools. They worked for simple feeds but couldn\u2019t handle custom fields, SEO metadata, or platform-specific formatting rules. WordPress plugins existed for syndication, but he had already migrated to EmDash for its security sandbox and modern stack. He needed a solution native to his new CMS.<\/p>\n
EmDash ships with a Model Context Protocol (MCP) server out of the box. That means any MCP-compatible AI agent<\/a> \u2014 Claude, Cursor, GitHub Copilot \u2014 can connect directly to the CMS and manage content programmatically. The developer used Claude\u2019s desktop app with the MCP integration to build a custom plugin.<\/p>\n The plugin registers a hook on the The plugin declares its capabilities in a manifest:<\/p>\n “`<\/p>\n capabilities: {<\/p>\n “read:content”: true,<\/p>\n “create:external”: [“medium”, “devto”, “linkedin”]<\/p>\n }<\/p>\n “`<\/p>\n Because EmDash runs every plugin inside a V8 isolate using Cloudflare\u2019s dynamic workers, this plugin cannot access the database, the file system, or any other part of the site. It can only read published content and make outbound API calls to the three declared endpoints. No other network access is permitted.<\/p>\n The developer tested the plugin in a local Node.js environment first, then deployed it to Cloudflare\u2019s paid plan where sandboxed execution is available. The total build time, including agent-assisted debugging, was about four hours.<\/p>\n The developer tracked his syndication time over two weeks. Before automation, syndicating a single post across three platforms averaged 15 minutes per platform \u2014 45 minutes total. After automation, the entire process took under 90 seconds: publish the post, wait for the webhook to fire, and verify the cross-post appeared on each platform. The plugin also preserved image alt text and canonical links, which his manual workflow frequently missed.<\/p>\n He also noted a reduction in formatting errors. Previously, he would catch one or two broken image references per week. In the two weeks after automation, he caught zero. The plugin uses the same portable text renderer for all outputs, eliminating the translation mismatches that happen when copying HTML between editors.<\/p>\n One caveat: the initial implementation hit API rate limits on Medium. Medium\u2019s API allows only one post per minute from a single token. The developer added a simple queue with a 70-second delay between posts. That extended the total syndication time to about four minutes for three platforms, still an 11x improvement.<\/p>\n The developer initially found the capability manifest restrictive. His first version tried to use a catch-all He also learned that MCP-based development changes the debugging cycle. Instead of writing code, testing, and deploying, he described the desired behavior to the agent, examined the generated code, and iterated through conversation. The agent handled the boilerplate for API authentication and error handling. His role shifted to specifying constraints and verifying outputs.<\/p>\n The biggest surprise was how easy it was to extend. After the basic syndication worked, he asked the agent to add a \u201ccross-post with custom excerpt\u201d feature for LinkedIn. The agent generated the change in under five minutes, and the dynamic worker sandbox validated that the new capability declaration matched the actual network calls.<\/p>\n EmDash\u2019s MCP server is not a bolt-on feature. It is part of the core architecture, alongside the plugin sandbox and the Astro-based theming engine. That means any MCP-compatible agent can manage content, create plugins, or migrate themes without manual coding. The syndication use case is one of dozens.<\/p>\n For agencies running multiple client sites, this pattern scales. An agent can be configured to syndicate content from a central EmDash instance to multiple destinations, each with its own formatting rules and API tokens. The sandbox ensures that a compromised plugin on one client site cannot affect another.<\/p>\n Still, the approach has limits. The developer noted that platform APIs change. When Medium updates its API, the plugin must be updated. Because the plugin is sandboxed, the update process is the same as the initial build \u2014 describe the change to the agent, regenerate the capability manifest, and redeploy. No core CMS update is required.<\/p>\n The developer\u2019s next project is a cross-CM syndication network: an EmDash plugin that pushes content to other EmDash instances, creating a decentralized publishing mesh. The MCP server makes that possible without building a custom API layer.<\/p>\n","protected":false},"excerpt":{"rendered":" The developer had a simple problem. Every time he published a post on his EmDash CMS blog, he had to manually copy it to Medium, Dev.to, and LinkedIn. Three platforms, three browser tabs, three rounds of formatting fixes. A 1,500-word article took 45 minutes to syndicate. He built a solution using EmDash\u2019s built-in MCP server. […]<\/p>\n","protected":false},"author":7,"featured_media":99260,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"fifu_image_url":"https:\/\/cards.overcentral.com\/cards\/en\/98021.png","fifu_image_alt":"Automating Content Syndication with EmDash\u2019s MCP Server","footnotes":""},"categories":[31],"tags":[],"class_list":["post-98021","post","type-post","status-publish","format-standard","has-post-thumbnail","category-technology"],"fifu_image_url":"https:\/\/cards.overcentral.com\/cards\/en\/98021.png","fifu_image_alt":"Automating Content Syndication with EmDash\u2019s MCP Server","_links":{"self":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/posts\/98021","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/users\/7"}],"replies":[{"embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/comments?post=98021"}],"version-history":[{"count":1,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/posts\/98021\/revisions"}],"predecessor-version":[{"id":99261,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/posts\/98021\/revisions\/99261"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/media\/99260"}],"wp:attachment":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/media?parent=98021"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/categories?post=98021"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/tags?post=98021"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}post:published<\/code>codecodecode event. When triggered, it reads the published post\u2019s portable text (EmDash\u2019s structured JSON content format) and maps it to each platform\u2019s API schema. For Medium, it sends a markdown conversion via the Medium API<\/a>. For Dev.to, it uses the Dev.to API with<\/a> tags and canonical URL. For LinkedIn, it posts a summary with a link.<\/p>\nThe Results: From 45 Minutes to 90 Seconds<\/h2>\n
Lessons Learned: Sandboxing Is a Feature, Not a Friction<\/h2>\n
network: \"*\"<\/code>codecodecode permission, which the dynamic worker rejected. He had to explicitly list each target API. That forced him to audit exactly what data his plugin was sending where. In practice, that audit saved him from accidentally exposing unpublished drafts \u2014 the plugin only receives published content.<\/p>\nWhy This Matters Beyond One Developer\u2019s Workflow<\/h2>\n