Agents: MCP and llms.txt

Coding agents can find, install and set up Chartory charts on their own. Connect the MCP server, or point the agent at llms.txt and AGENTS.md, then ask for a chart in plain words.

MCP server

The server lets your agent search charts, get the code, the spec and data schema, and the theme tokens. Search, specs and tokens answer without a token; the code of the paid types uses your registry token. The server is in the official MCP Registry as com.usechartory/chartory, and github.com/chartory/charts is also a Cursor plugin.

One command adds the server.

claude mcp add --transport http chartory https://usechartory.com/mcp \  --header "Authorization: Bearer $CHARTORY_TOKEN"

other clients · mcp config

"chartory": {  "url": "https://usechartory.com/mcp",  "headers": { "Authorization": "Bearer ${CHARTORY_TOKEN}" }}

What the server can do

Ask forYou get

Search charts · search_chartsTypes that answer your question, with their slugs

The code · get_chart_codeThe component, ready for npx shadcn add

The spec · get_chart_specProps, sizes and the JSON Schema of the data

The tokens · get_tokensThe token names and the four modes

llms.txt

A plain-text map of the docs at usechartory.com/llms.txt, and the whole text in llms-full.txt. Every page also has a Markdown twin.

AGENTS.md

Drop it into your repo so agents follow the rules: use the tokens, never scale a chart, pick a size that fits the card.

Prompts that work

Add a waterfall of monthly cash flow to the finance page, size M.
Show order book depth for BTC in the trading card, size S, no legend.
Replace the bar chart in the uptime card with uptime bars.