Conciseness

The body is lean and assumes Claude's competence, opening with a single orienting line and then tight command sections; it avoids explaining what a CLI or sandbox is and every token earns its place.

Actionability

Provides concrete, copy-paste-ready bash commands for login, headless runs, sessions, worktrees, and diagnostics, covering the common cases with exact flags.

Workflow Clarity

Sequences local-evidence inspection before advice, distinguishes permission vs sandbox controls, and includes explicit validation/safety checkpoints (dry-run before worktree deletion, verifying destructive commands) with feedback-loop guidance for risky operations.

Progressive Disclosure

SKILL.md is a concise overview with two clearly signaled, one-level-deep references ([references/cli-reference.md], [references/workflows.md]) that exist in the bundle; bulk detail is split out and navigation is explicit.

Specificity

Enumerates concrete actions and surface area — interactive/headless runs, authentication, sessions, worktrees, permissions, sandboxing, MCP servers, plugins, inspection, updates, automation output — rather than vague abstractions.

Completeness

Explicitly answers both 'what' (reference/workflow guidance for the grok CLI) and 'when' ('Invoke only when the user explicitly requests...') with concrete trigger guidance and explicit anti-triggers.

Trigger Term Quality

Includes natural trigger phrases ('use Grok Build CLI', 'grok-build skill', 'workflow requires Grok Build') but is lighter on common synonyms/extensions a user might casually say; coverage is good but not exhaustive.

Distinctiveness Conflict Risk

Scopes tightly to the Grok Build CLI and explicitly excludes general coding, generic shell tasks, and other agent CLIs, giving it a clear niche with minimal conflict risk.