logs_tail

Follows new log rows as pages send them: a long poll that returns as soon as rows arrive, with a cursor for the next call.

These tools read ParallelSandbox's own log service, which pages feed with the browser SDK; they do not read Sentry or any other tracker. A log project is created over REST, not in the app: POST https://log.parallelsandbox.com/v1/projects with the account's token (an API key or an OAuth access token) returns its id prj_... and a write key psw_....

• project: the project id prj_... or its name. A name the account does not have fails with project not found, followed by the projects the account does have (name (prj_...)), or how to create one when there are none. There is no project for ParallelSandbox's own services, nor for your servers and containers: read those in the box (docker logs, the logPath of a background command).
• since: a duration (15m, 2h, 3d, 1w), an RFC3339 timestamp or unix time; default 1h.
• query: plain text matches as a whole term; text starting with {, ?, -, ", % or [ is passed through as a CloudWatch filter pattern, such as { $.level = "warn" } or ?timeout ?retry.
• boxId: only logs sent from pages served by that box.
• limit: rows per page, default 200, at most 1000; pass nextCursor back as cursor for the next page.
• logs_tail is a long poll: it returns as soon as new rows arrive or after waitSec (default 20, at most 25), with a cursor to pass to the next call; omit cursor to start from now.
• Each row has ts, ingestedAt, level, box, release, platform, route, msg, stack, stackResolved, extra and stream. Logs are kept without expiry.

Every tool and topic is listed in the tool reference.