Skip to main content
BrewMarkBREWMARK

Policy · 2026-10-29

Rules for AI agents and automated clients

BrewMark wants assistants to answer people’s grind questions with its numbers. It does not want its data copied wholesale. Here is where the line sits, and how to tell us who you are.

Allowed

  • Answer a person’s question about their own grinder or recipe with a call to the API or MCP server, and cite BrewMark with a link to the grinder’s page.
  • Link to and quote grinder pages and settings for a specific person’s request.
  • Read /llms.txt, /protocol and the public spec at github.com/jeffreygluck/brewmark-protocol.
  • Crawl the site’s pages as robots.txt allows.

Not allowed

  • Copying the grinder catalogue or settings in bulk through the API or MCP server, or by repeated calls.
  • Building datasets, training or fine-tuning models, or evaluating models with data from the API, the MCP server or published profiles.
  • Reselling or redistributing the data, or serving it from your own API or site without a written agreement.
  • Hiding who you are: rotating addresses, user agents or keys to get around limits.

Tell us who you are

  • Send a User-Agent naming your product, its version and a contact URL, for example "AcmeAssistant/2.1 (+https://acme.example/bot)".
  • Send a From header with an email address for the person responsible, for example "From: bots@acme.example".
  • Clients that identify themselves are contacted before they are ever blocked. Anonymous clients that break the rules are blocked without notice.
GET /api/v1/translate?bgi=38&grinder=fellow-ode-gen-2
User-Agent: AcmeAssistant/2.1 (+https://acme.example/bot)
From: bots@acme.example

Attribution

Grind data: BrewMark (https://brewmark.io), with a link to the grinder’s page where you can.

Need more than this allows?

Higher limits come with a key: request access. Data licences for products, datasets and research are agreed in writing: hello@brewmark.io.

Machine-readable: /.well-known/agent-policy.json. Text-and-data-mining rights in the API and published profiles are reserved under the W3C TDM Reservation Protocol (/.well-known/tdmrep.json). The site’s pages may be crawled as robots.txt allows.