grok-skills

Use when the user says to use Grok, wants a Grok review, wants a second opinion from Grok, or asks to delegate a task to Grok. Also use for getting an independent code review from Grok, having Grok reason about an architecture or debugging question, having Grok implement code (live tree by default, or opt-in isolated worktree), or having Grok independently verify someone else's implementation. Trigger phrases include "use grok", "grok review", "second opinion from grok", and "delegate to grok".

Check Grok readiness and optionally toggle stop gate / run mode / Codex agents scope (Codex agents auto-install on SessionStart)

Have Grok implement code (default: live tree; opt-in isolated worktree). Nothing is committed or pushed

One-call delegate: Grok code then auto-handoff verification (verify-only; mode-aware integrate elsewhere)

ACP multi-turn peer channel (start/prompt/stop). Default peer path; opt out with GROK_DISABLE_ACP=1

Read a verified implementation handoff for a completed Grok code run (runId only; no apply)

Report (dry-run) or remove a Grok run's owned session state and worktree by run id

Cancel a running Grok job

Show stored output for a finished Grok job

Read a Grok run by id: finished envelope, or running progress if still in progress (read-only)

Get a cold, isolated Grok second opinion on named artifacts (architecture, debugging, plan critique)

Run a full-context, read-only Grok code review over a workspace path

Have Grok independently verify a change in an existing worktree (no source edits, hermetic)

Run adversarial-review then review on the same target (dual-lens harden recipe)

Hostile Grok review that challenges design; web on by default

Check Grok wrapper readiness (binary, auth, sandbox, state) - no task is run

Bounded Grok debate: two opposing reason passes, then synthesis

List recent Grok jobs for this workspace

Package the current Claude session into a Grok task pack