Close study · not a fake MCP listing
SKILL.md — Grok
Real skill file for agents. When to use / when not is inside the document. This is not a fake MCP listing.
File
Raw
--- name: grok description: Operate Grok as an agent using public APIs, CLIs, and docs. Twin Score agent 70. --- # Grok skill Grok is xAI's assistant, known for real-time flavor from X and a less filtered tone than the larger labs. The human app and the Grok experience inside X are the main surfaces. API access exists but documentation, SLA language, and enterprise packaging lag Claude and OpenAI, which is why the agent Twin Score trails the human score. We review the public product, not the meme. Twin Score on this directory is 74 for humans and 70 for agents, using the public rubric (interface, onboarding, human docs, collaboration, support, pricing fairness, accessibility, reliability versus API quality, MCP or tool surface, CLI, structured docs, machine auth, quotas, determinism, and skill packaging). Grok is sold by xAI, founded 2023, headquartered in San Francisco, USA. Pricing in public materials: Consumer access bundled with X Premium tiers; API available from xAI. Machine capabilities we record: API strong, MCP none, CLI basic, docs basic, auth xAI API keys. This page is editorial, not a star rating scraped from users. ## When to use Use this skill when a user asks to integrate, script, or evaluate Grok (xAI). ## When not to use Do not claim an MCP server exists unless the capability matrix says so (current MCP level: none). Do not invent user-count ratings. ## Capability matrix | Surface | Level | |---|---| | API | strong | | MCP | none | | CLI | basic | | Docs | basic | | Auth | xAI API keys | | Webhooks | strong | | SDKs | http, python | ## Auth xAI API keys ## Pricing (public) Consumer access bundled with X Premium tiers; API available from xAI. ## Official site https://x.ai ## Twin Score Human 74. Agent 70. Updated 2026-08-01. Rubric: https://buildsomethingagentswant.com/scorecard ## Alternatives - chatgpt - claude - gemini