# Machine-readable access

> Append /md to any Lacuna URL and an agent gets the page as clean markdown. The schema lives at llms.txt.

Agents are among Lacuna's readers, so every content page has a machine door. The full schema is documented at [llms.txt](https://lacuna.tiptreesystems.com/llms.txt).

| Surface | What it returns |
| --- | --- |
| /md on any content page | The page as text/markdown, no HTML; research directions, papers, author profiles, and proposals, typically 5–30 KB each |
| /direction/…/csv | A data export per research direction |
| /figures/… | The extracted figures, as PNG or JPEG |
| /sitemap.xml, /llms.txt | The crawl map and the schema notes |
| MCP server | Lacuna as a research-context provider for agent runtimes; search papers, inspect author and direction context, pull source-grounded summaries |

The scope note in llms.txt is worth quoting: "Content is generated from extracted concepts, not raw paper text."

## Use it from your own tools

You do not need Althea to build on the commons. Point an agent runtime or a local coding environment at the MCP server, or fetch the /md pages directly from whatever you already run; either way, your tools inherit the same source links a human reader gets. The MCP package and install command depend on your release channel, so check the release notes or your workspace setup before wiring it in.

## Citations are the front door

When someone asks their assistant a question and the citation that comes back is a Lacuna page, that citation is how they find the commons. Machine-readable pages keep those citations cheap to produce and easy to check. This hub follows the same rule. Any page here can be copied as Markdown.