Get markdown for a model | Stophy

Every endpoint can answer with markdown instead of JSON. Use it when a model reads the result. Use JSON when your code reads it.

Ask for markdown

Send accept: text/markdown:

curl -X POST https://api.stophy.dev/v1/web/search \
  -H "content-type: application/json" \
  -H "accept: text/markdown" \
  -d '{"query":"bun runtime","limit":2}'

This is the response to that request. Web search works without a key, so you can run it as is.

**web.search**: 2 results

## Results

| Title | Description | Domain |
| --- | --- | --- |
| [Bun — A fast all-in-one JavaScript runtime](https://bun.com/) | Bundle, install, and run JavaScript & TypeScript — all in Bun. Bun is a fast JavaScript runtime & toolkit with a bundler, test runner, and npm-compatible package manager built in. | bun.com |
| [GitHub - oven-sh/bun: Incredibly fast JavaScript runtime, bundler, test runner, and package manager – all in one](https://github.com/oven-sh/bun) | 3 days ago - At its core is the Bun runtime, a fast JavaScript runtime designed as a drop-in replacement for Node.js. | github.com |

What you get

The first line names the endpoint and how many results it returned.
Short rows become tables. Longer items get their own heading.
Links stay links, so the model can cite its sources.
Empty values are left out.
When there is another page, the last line says which cursor to send. See Page through results.

Headers and errors

The response is content-type: text/markdown; charset=utf-8.
What the call cost is in the x-credits-used header.
Errors are still JSON, with the same shape as always. See Errors.

Over MCP, results are markdown by default. Pass format: "json" to stophy_call for JSON.