deepline feedback

Usage: deepline feedback [options] [command]

Submit CLI feedback to Deepline.

Options:
  -h, --help             display help for command

Commands:
  send [options] [text]  Send CLI feedback to Deepline.
  help [command]         display help for command

Usage: deepline feedback send [options] [text]

Send CLI feedback to Deepline.

Arguments:
  text                       Feedback text

Options:
  --file <path>              Read feedback text from a file, or - for stdin
  --command <command>        Command that reproduced the issue
  --payload <payload>        JSON or plain-text payload for the repro
  --error-outcome <outcome>  Error impact on the task: terminal or continued
                             (choices: "terminal", "continued")
  --requested                Mark the report as explicitly requested by the user
  --json                     Emit JSON output
  -h, --help                 display help for command

Explanation:
  Purpose:
    Send a bug report or product feedback to the Deepline team. The calling
    agent's session id is attached, so a report can be matched to a transcript.
  When:
    The user asks to report something: confirm the text with them first, pass
    --requested, and offer to share the session with sessions send.
    Unasked, when you see friction Deepline could act on, unless the user
    restricts external reporting: a play far slower than its input size,
    earlier comparable runs, or visible progress would suggest, or one that
    looks stuck; or an ordinary result that took avoidable discovery,
    conversion, id extraction, retries, exports, or a manual workaround. Send
    one concise report per issue, without --requested and without a
    transcript, then keep working on the task. Provider work that is
    legitimately long is not friction.
  Input:
    Positional text or --file <path> (- for stdin). --command "<the command
    that reproduced it>" and --payload <json or text> attach a repro.
    --error-outcome terminal when an error stopped the task, continued when
    work went on, including through a workaround; omit it when no error
    occurred. --requested only when the user asked for the report. --json.
    Put the evidence in the text. A slow run: the goal, the play and run id,
    input size, observed duration, status changes or polling count, and what
    you expected. Extra steps: the goal, the command path you took, the
    detours, and the direct path you expected.
  Output (JSON):
    A confirmation envelope for the report.
  Cost:
    Free. A report does not authorize a paid reproduction or a session upload.