Reblog developer documentation
Everything Reblog exposes to code: a REST Content API to read your published articles, and an MCP server that lets an AI agent run the blog end to end.
Reblog is a headless blogging platform: you write, translate and publish inside the dashboard, and your site reads the result. Two surfaces are open to code, and they answer two different questions.
Content API
REST"Give me my published articles so my site can render them." One authenticated GET endpoint: a single article by handle, a paginated category feed, or the full sitemap.
MCP server
56 tools"Let an AI agent run the blog." Write, translate, illustrate, publish, schedule and audit articles from Claude, Claude Code, Cursor, or anything else that speaks MCP.
Start here#
API overview
How the whole thing fits together: base URLs, the two authentication paths, and the full list of what a connection can do.
Authentication
Create a project API key, pick a permission level, and send it correctly.
Connect a client
Copy-paste setup for Claude, Claude Code, Cursor and every other MCP client.
Tool reference
All 56 MCP tools, grouped by the permission that unlocks them, with their arguments.
Permissions
Scopes, consent presets, project roles, and how the two are intersected on every call.
OAuth 2.1
For app builders: discovery, dynamic client registration, PKCE, tokens and revocation.
Sixty seconds in#
- 1
Create a project API key
In the dashboard, open your project and go to API keys. Choose Read-only if the key only feeds a website. The private key is shown once.
- 2
Read your articles
One request returns a published article, body included.
bashcurl -H "Authorization: Bearer $REBLOG_API_KEY" \ "https://reblog.so/api/external/articles?handle=guides/getting-started" - 3
Or hand the blog to an agent
Point any MCP client at the server URL and approve the access level you want it to have.
bashclaude mcp add --transport http reblog https://reblog.so/api/mcp
Everything here is public
These pages describe endpoints that are open to any customer of Reblog. They do not cover the dashboard's own internal endpoints, which require a browser session and are not part of the supported surface.