https://nicolevanderhoeven.com/tags/cli/Recent content in Cli on Nicole van der HoevenHugo -- gohugo.ioen-usNicole van der HoevenWed, 03 Jun 2026 18:00:00 +0100https://nicolevanderhoeven.com/blog/20260603-context-window-04-cli-mcp/Wed, 03 Jun 2026 18:00:00 +0100https://nicolevanderhoeven.com/blog/20260603-context-window-04-cli-mcp/<p>MCP vs CLI: which one should your AI agent actually use? In this episode of <em>The Context Window</em> — Grafana&rsquo;s livestream series on AI in observability — I&rsquo;m joined by <a href="https://www.linkedin.com/in/tiffanyjerniganm/">Tiffany Jernigan</a> and three of the engineers and PMs behind Grafana&rsquo;s agent tooling: Ben Sully (who wrote the Grafana MCP server), and Ward Bekker and Dafydd Thomas from the team building <code>gcx</code>. We get into the two main ways an AI agent can reach the outside world — the Model Context Protocol (MCP) and good old command-line interfaces — and look at when each one makes sense, the tradeoffs around context windows, tool sprawl, token cost, and reliability. We walk through Grafana&rsquo;s cloud and open-source MCP servers along with <code>gcx</code>, the Grafana Cloud CLI built for agents, do a live side-by-side demo of both creating an SLO, and talk about how we benchmark them against each other with o11y-bench.</p> <div style="position: relative; padding-bottom: 56.25%; height: 0; overflow: hidden;"> <iframe allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share; fullscreen" loading="eager" referrerpolicy="strict-origin-when-cross-origin" src="https://www.youtube.com/embed/ZR6U4TN8-Cw?autoplay=0&amp;controls=1&amp;end=0&amp;loop=0&amp;mute=0&amp;start=0" style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; border:0;" title="YouTube video"></iframe> </div> <h2 id="timestamps">Timestamps</h2> <ul> <li>00:00:00 — Introductions</li> <li>00:02:49 — The last month in AI news</li> <li>00:10:19 — What is Grafana Assistant?</li> <li>00:12:30 — What does it mean for Assistant to use tools?</li> <li>00:13:30 — What is MCP?</li> <li>00:16:56 — What is a CLI in an AI context?</li> <li>00:19:17 — What are skills?</li> <li>00:24:37 — Can you use a custom MCP server with Grafana Assistant?</li> <li>00:25:55 — Why team MCP?</li> <li>00:28:22 — The context-window cost of MCP tool definitions</li> <li>00:31:50 — Tool discoverability</li> <li>00:39:25 — What is gcx, and why a CLI?</li> <li>00:46:13 — Demo: creating an SLO with MCP vs gcx</li> <li>00:52:58 — Benchmarking MCP vs gcx with o11y-bench</li> <li>00:58:10 — How MCP and gcx relate to Grafana Assistant</li> <li>01:02:35 — k6 MCP vs Grafana MCP</li> <li>01:03:00 — When to use which (decision guide)</li> <li>01:04:29 — Adding the OSS Grafana MCP to Assistant</li> <li>01:05:41 — Turtles all the way down</li> </ul> <h2 id="resources">Resources</h2> <h3 id="mcp-and-cli-for-agents">MCP and CLI for agents</h3> <ul> <li><a href="https://modelcontextprotocol.io">Model Context Protocol</a></li> <li><a href="https://github.com/grafana/mcp-grafana">Grafana MCP server (OSS)</a></li> <li><a href="https://github.com/grafana/gcx">gcx — the Grafana Cloud CLI</a></li> <li><a href="https://grafana.com/blog/o11y-bench-open-benchmark-for-observability-agents/">Introducing o11y-bench: an open benchmark for observability agents</a></li> <li><a href="https://slack.grafana.com">Grafana Community Slack</a></li> </ul> <h3 id="related-reading">Related reading</h3> <ul> <li><a href="https://contexthorizon.substack.com/p/unlearnings-from-building-grafana">My Context Horizon piece: <em>Unlearnings from building Grafana Assistant</em></a></li> </ul> <h3 id="more-from-the-context-window">More from The Context Window</h3> <ul> <li><a href="https://nicolevanderhoeven.com/blog/20260422-context-window-03-ai-observability/">#03: All About AI Observability</a></li> </ul>