Skip to content
NativPost API

MCP overview

The NativPost MCP server exposes this API as tools for AI assistants. Instead of writing HTTP calls, you talk to your workspace:

“What’s my brand voice, and which platforms am I connected to?”

“Draft three Instagram captions about our new pricing and save them as drafts.”

“What’s scheduled for next week?”

It speaks the Model Context Protocol — an open standard for connecting AI assistants to external systems — so it works with Claude Desktop, Claude Code, and any other MCP client.

What you can do with it

Section titled “What you can do with it”

The server covers the same surface as the REST API: 20 tools across workspace, content and campaigns. Anything you can do with a curl call in these docs, an assistant can do through a tool.

AreaTools
Workspaceaccount, brand profile (read + write), social accounts, media library, templates, analytics, webhooks
Contentlist, get, create, update, delete, publish
Campaignslist, get, create, update, delete, launch

See the Tool reference for every tool and its arguments.

How it relates to the REST API

Section titled “How it relates to the REST API”

The MCP server is a client of this API — it holds your API key and makes the same authenticated requests documented here. That means:

  • Same permissions. A key that can’t reach an endpoint can’t reach it through MCP either.
  • Same limits. Plan quotas and rate limits apply identically.
  • Same errors. Failures come back carrying the API’s own code and message, so the assistant can explain what went wrong.

If you’re building an application, use the REST API directly. If you want a person to work with their content conversationally, use MCP.

Safety

Section titled “Safety”

Tools are annotated so MCP clients can tell reads from writes:

  • nativpost_delete_content and nativpost_delete_campaign are marked destructive.
  • nativpost_publish_content and nativpost_launch_campaign are marked open-world — they reach live social accounts and consume quota.

Well-behaved clients ask for confirmation before running those. Treat an API key given to an assistant as one that can post publicly on your behalf, and scope it accordingly.

Next

Section titled “Next”