Skip to main content
WPOS is built to be driven by agents, not just clicked through. The way you access it programmatically today is the Model Context Protocol (MCP), plus these docs served in agent-readable Markdown. That means your own Claude or Cursor can read, reason about, and operate WPOS directly.
WPOS does not publish a general-purpose public REST API yet. The supported way to build on WPOS today is MCP. If you have a programmatic need that MCP does not cover, tell us at support@wpos.ai.

​
The docs MCP server

These docs are served as an MCP server, so any MCP-capable agent can load them and answer questions or operate WPOS with full knowledge of how it works. 
https://docs.wpos.ai/mcp
Add the server to your client, and the agent can cite these pages directly while helping you operate sites. See MCP for the bigger picture of how MCP shows up across WPOS.

​
Markdown for agents

Every page here is also plain Markdown, which keeps token usage low for any agent reading the docs.

Per-page Markdown

Add .md to any docs URL to get the raw Markdown, for example docs.wpos.ai/get-started/quickstart.md.

Whole-site index

/llms.txt lists every page; /llms-full.txt is the full corpus in one file.

​
Operating WPOS programmatically

Once your agent is connected, it operates WPOS the same way the Command Center does, and against the same guardrails:
If your team already lives in Claude or Cursor, this is the fastest way to fold WPOS into your workflow. No new API client to build, just point your agent at the MCP server.

​
When a REST API arrives

When WPOS publishes a public REST API, this section becomes a full interactive reference: drop an OpenAPI spec into the repo, point docs.json at it, and Mintlify generates a playground per endpoint. Until then, MCP is the supported and documented surface.