API

Base URL

https://services.my-beast.com/logs

Send every request over HTTPS. The API has one endpoint, which takes a batch of lines. Reading lines back is done on your project's pages: Search and Live.

Authentication

Send the project's server secret for Logs as a bearer token: Authorization: Bearer sk_…. It is the same secret as MYBEAST_LOGS_SECRET, found or made as the quickstart says. A Logs key opens Logs only, and its project is the one the lines go to.

Send lines

POST/v1/logs

The body is {"lines": [...]}, at most 1,000 lines and 4 MiB as sent, with Content-Type: application/json. Gzip it to send more at once, with Content-Encoding: gzip: unpacked, it may be up to 32 MiB. The Laravel SDK always does.

{
  "lines": [
    {
      "id": "01JBFX3W2M4R8Y6T0Q9N7K5H3D",
      "at": "2026-10-27T12:34:56.123456+00:00",
      "level": "info",
      "message": "Invoice 1042 sent",
      "context": { "invoice": 1042 },
      "channel": "app",
      "environment": "production",
      "release": "9c1e4b2",
      "host": "web-1",
      "request_id": "01JBFX3W2KQ7S3VZ0H5M9N2P4R",
      "transaction": "POST /invoices/{invoice}/send"
    }
  ]
}

curl -X POST https://services.my-beast.com/logs/v1/logs \
  -H "Authorization: Bearer $MYBEAST_LOGS_SECRET" \
  -H "Content-Type: application/json" \
  -d @batch.json

gzip -c batch.json > batch.json.gz
curl -X POST https://services.my-beast.com/logs/v1/logs \
  -H "Authorization: Bearer $MYBEAST_LOGS_SECRET" \
  -H "Content-Type: application/json" \
  -H "Content-Encoding: gzip" \
  --data-binary @batch.json.gz

{
  "accepted": 1,
  "duplicates": 0,
  "refused": []
}

Scrub personal data before you send it: Logs scrubs every line again on arrival, but only with its own patterns. See Data and privacy.

A line

Each line in lines is an object with these fields:

Each line stands on its own: one that doesn't fit is refused and named, and the rest go in. What is only too long is cut, never refused.

The answer

A batch that is read answers 202 Accepted with three things:

• accepted: how many lines were stored.
• duplicates: how many the project had already, by their id, from a batch sent before. They aren't stored or counted again.
• refused: the lines that weren't taken, each with its id and a reason.

{
  "accepted": 1,
  "duplicates": 0,
  "refused": [
    { "id": "01JBFX3W2N5S9Z7V1R0P8M6J4E", "reason": "at: older than the 14 days this environment keeps lines" },
    { "id": "01JBFX3W2P6T0A8W2S1Q9N7K5F", "reason": "monthly-limit" }
  ]
}

A reason is a sentence that names the field at fault, for you to read, except monthly-limit: the project has used up its lines for this month, and the rest of the month's are refused until the 1st, UTC. See Lines a month. A refused line will be refused again, so don't send it twice.

Errors

When the whole batch isn't taken, the status says why and nothing of it is stored:

A 401, 402, 413 or 422 comes with JSON that has a message. A 402 adds a type and a link to the page where the subscription is renewed:

{
  "message": "unauthenticated"
}

{
  "message": "Shop's subscription ended on 1.12.2026: nothing new is taken in or added. Renew it at id.",
  "type": "subscription_ended",
  "link": "https://id.my-beast.com/acme/-/subscription"
}

Retrying

• 429, 5xx and network errors: keep the lines and send the same batch again later, waiting longer after each failure. Their ids make it safe: a line stored the first time comes back as a duplicate.
• 401 and 402: keep the lines, fix the secret or renew the subscription, then send them.
• 413: send the same lines in smaller batches.
• 422: the same body will fail again. Set it aside, so it doesn't hold up the lines after it, and fix what makes it.

The Laravel SDK does all of this for you. How lines are sent says how.