/ask

Ask Grok a one-shot question.

Raw slash-command arguments: $ARGUMENTS

Execution mode:

• If the raw arguments include --background, use the background flow.
• Otherwise use the foreground flow.

Foreground flow:

node "${CLAUDE_PLUGIN_ROOT}/scripts/grok.mjs" ask "$ARGUMENTS"

• Return the command stdout verbatim — do not summarize, paraphrase, or add commentary before or after it.
• Preserve any inline markdown citation links Grok includes.

Background flow:

• Launch with Bash in the background:

Bash({
  command: `node "${CLAUDE_PLUGIN_ROOT}/scripts/grok.mjs" ask "$ARGUMENTS"`,
  description: "Grok ask",
  run_in_background: true
})

• Do not wait for completion in this turn. The command prints a job id immediately; tell the user: "Grok ask started in the background. Check /grok:status for progress, then /grok:result <job-id> for the answer."

Notes:

• Grok answers with live web/X search on by default (the main reason to reach for Grok mid-task), following the web_search config setting. Pass --no-search to disable it for one call, or --search to force it on.
• The default model is the configured search_model (grok-4.5). --model accepts only grok-4.5 or grok-composer-2.5-fast; legacy IDs such as grok-build are deprecated and rejected.
• --max-turns <n> overrides the configured max_turns for this call. --effort and --reasoning-effort are accepted as aliases and sent to Grok as one canonical --reasoning-effort argument (the long form wins if both are given).
• --thought includes Grok's returned reasoning in a collapsed details block.
• --print-args prints the exact Grok CLI argv and exits without calling Grok; it takes precedence over --background.
• This command is read-only — it only asks Grok a question and prints the answer.