check

Check the status of one or all ebb-ai deferred tasks.

Arguments

$ARGUMENTS

• If $ARGUMENTS looks like a single task id (alphanumeric, dash, underscore only), check that one specific task.
• If $ARGUMENTS is empty, list every known task — the summary always includes queued, scheduled, running, completed, failed and cancelled tasks (there is no age cutoff), including tasks persisted by previous sessions.
• --all / all is accepted for compatibility and behaves the same as an empty argument.

What to do

1. Call the ebb-ai MCP server's check_queue_status tool with:
   • task_id — the parsed id, omit for the list view.
2. Render the result as a compact table — never raw JSON. For each task:
   • id (last 8 chars suffice)
   • status — queued, scheduled, running, completed, failed, cancelled
   • scheduled_for — relative time ("in 2h", "yesterday 14:30")
   • region
   • For completed tasks: append the carbon receipt (fields as they appear in the response, when present):
     • estimated_carbon_g (what the scheduler projected)
     • actual_carbon_g (what was billed against the actual grid intensity at dispatch time, if present)
     • delta_pct — +X% or -X% actual vs estimated
     • grid_source — which feed produced the intensity. If it is mock the response says so explicitly ("SYNTHETIC (mock) grid data") — repeat that caveat instead of presenting the grams as measured.
     • energy_source — confidence tier of the per-model energy coefficients (measured / estimated / fallback)
   • For failed: short error reason.
3. If exactly one task is requested and the response carries a Result: block (the LLM response), include it verbatim under the table.
4. End with a short hint line:
   • If anything is queued or scheduled: "Next dispatch tick: ~5 min (if the ebb tick daemon is installed — otherwise run ebb tick --once)."
   • If everything is completed: nothing.