Grafana Cloud CLI (gcx): Terminal-Based Observability for Developers and AI Agents
Introduction
The way software is built is evolving rapidly, and with it, the methods teams use to monitor and troubleshoot their systems must evolve too. Modern engineering workflows increasingly rely on command-line interfaces (CLIs), especially as AI-powered coding assistants like Cursor and Claude Code handle a growing share of routine development tasks. These tools accelerate code generation, but they don't eliminate the friction of switching contexts to check dashboards or investigate incidents. More critically, they create a new blind spot: agents can see your codebase but remain unaware of what's actually happening in production. They don't detect a sudden latency spike, know if service-level objectives (SLOs) are being met, or understand real-time system behavior. Instead, they operate on assumptions rather than live data.
The Changing Development Landscape
Engineers now spend a significant portion of their day inside a terminal, often working alongside agents that automate code writing, testing, and refactoring. While this boosts productivity, it introduces a disconnect. When an issue arises, the typical response involves jumping out of the terminal into a separate monitoring tool — breaking flow and slowing down resolution. Moreover, agents that lack production context cannot prioritize fixes, adjust code based on real metrics, or validate that changes actually improve performance. Bridging that visibility gap is essential for keeping pace with faster development cycles.
Introducing gcx: Grafana Cloud CLI
To address these challenges, Grafana has released the public preview of gcx, a new command-line tool that brings Grafana Cloud and its AI assistant directly into your terminal. With gcx, developers and their agents can observe, alert, and respond to incidents without leaving the command line — reducing reaction time from hours to minutes.
From Greenfield to Full Observability in Minutes
Most services begin their lifecycle with zero instrumentation, no alerts, and no SLOs. That's the norm, not an exception. gcx treats this as a starting point rather than a blocker. You simply point your agent at the service and ask it to bring observability up to standard. The tool exposes all the primitives needed across the full observability lifecycle:
- Instrumentation: Wire OpenTelemetry directly into the codebase. Validate that metrics, logs, and traces are flowing. Confirm data lands in the correct backends — all from the terminal.
- Alerting, SLOs, and Synthetics: Generate alert rules based on the signals your service emits. Define an SLO using real latency or availability metrics and push it live. Deploy synthetic checks so users aren't the first to report an outage.
- Frontend and Backend Observability: Onboard a Faro-instrumented frontend, create the app, and manage sourcemaps for readable stack traces. Similarly, onboard backend services and Kubernetes infrastructure using Instrumentation Hub.
- Everything as Code: Pull dashboards, alerts, SLOs, and checks as local files. Edit them with your agent, then push changes back. When a human needs to investigate, open a deep link directly into Grafana Cloud.
With all these capabilities in one place, what traditionally required a multi-day ticket now becomes a single session with an agent.
Why This Matters for AI Agents
The true power of gcx emerges when agents gain access to it. Without production context, an agent merely pattern-matches on source code and hopes for the best. With gcx, the same agent can query the live system state — checking latency distributions, error rates, and SLO compliance — and make informed decisions based on actual data. This transforms an agent from a blind code generator into a context-aware assistant that understands real-world system behavior.
For example, when an agent proposes a performance improvement, it can first validate current latency metrics via gcx, then after deploying changes, verify that the improvement actually took effect. This closes the feedback loop and ensures every code change is grounded in observable reality.
Getting Started with gcx
To begin using gcx, install the CLI tool from Grafana's official repository. Authenticate with your Grafana Cloud account, and you're ready to instrument services, set up alerts, and manage dashboards — all from the command line. The tool seamlessly integrates with popular agent frameworks, enabling your AI co-pilot to act on real-time telemetry.
Internal anchor links: Introduction, Changing Landscape, Introducing gcx, Quick Onboarding, Why Agents Need This, Getting Started.
In summary, gcx bridges the gap between development and operations, bringing observability directly into the engineer's primary workspace — the terminal — and empowering AI agents with the production context they need to make smarter decisions.