Share AI agent output as a page

The answer was good. The screenshot of the answer is not.

# Chat is where documents go to die

An agent can produce a genuinely good document — a migration plan, an API walkthrough, a careful explanation — and then the artifact problem starts. Paste it into Slack and the formatting dies. Screenshot it and the text dies. Export it to a file and the link dies. The output was never the problem; the container was.

With lucid.page connected, the agent publishes the document itself, and the container is a web page: typeset, linkable, and stable.

# Formatting survives the trip

Because the page renders real Markdown, everything the agent wrote arrives intact — fenced code highlighted with a copy button, TeX math set properly, Mermaid fences drawn as diagrams:

$$L = -\sum_{i} y_i \log \hat{y}_i$$

A table stays a table. A heading stays a linkable anchor. The reader gets the document, not a photograph of it.

# Share the link, keep the source

The page is unlisted by default — it reaches exactly who you send it to. The Markdown source stays one URL away at /raw/, so the next agent in the chain can read it as text instead of pixels. And when the document needs a fix, the page updates in place; the link never changes.

# Common questions

# How do I get an agent to publish its output?

Connect the lucid.page MCP server — see the agent documentation — then ask. ‘Publish this as a page’ is enough; the agent calls the tool and returns the link.

# Do math and code survive?

Yes. Fenced code blocks are syntax-highlighted with a copy button, $...$ and

$$...$$

# Can the shared page be updated later?

Yes — the same agent, or any client with access, can update the page and the URL stays the same. Previous versions are kept as revisions.