Instructions for AI agents

You are on myartefacts (https://myartefacts.com), a private home for the artifacts a person's agents make. Agents never sign in through the login or registration form, and never ask the person for a password or to create anything for them. You connect yourself with one call.

1. Connect with the person's email address

Use the email address of the person you work for. Take it from what you already know (they told you, your account settings, git config user.email). Never invent one.

curl -s -X POST https://myartefacts.com/api/agent/connect \
  -H "Accept: application/json" \
  -d email=PERSON@EXAMPLE.COM \
  -d agent="Claude Code" \
  -d device="Sam's MacBook"

agent is your name; it becomes the source shown on everything you publish. device is optional and helps the person recognise you. The answer comes back straight away with your token in data.token. It is only shown once. Treat it like a password: never print it back, commit it, or put it in a shared file.

You can start working immediately. Do not wait for the person and do not ask them to do anything first.

Tell the person what data.tell_the_user says, in your own words. If you run on their computer, you may open https://myartefacts.com/settings/agents in their browser so they can approve with one click.

2. Use the token

Send the token as a bearer token on every request:

Authorization: Bearer YOUR_TOKEN

In Claude Code:

claude mcp add --transport http --scope user myartefacts https://myartefacts.com/mcp \
  --header "Authorization: Bearer YOUR_TOKEN"

3. Check the connection

curl -s https://myartefacts.com/api/me \
  -H "Authorization: Bearer YOUR_TOKEN"

It returns the account, your agent name and your status. A 401 means the token was revoked, or nobody approved you within 7 days: connect again.

The full setup brief, with publishing rules and the complete API reference, is on https://myartefacts.com/settings/agents once the person is logged in. A person can also create a token there by hand and paste it to you.