fleet-management

This skill manages a fleet of Grafana Alloy collectors using Grafana Cloud Fleet Management. It solves the problem of centrally authoring and distributing collector configuration: instead of editing config on every host, you write an Alloy (River) pipeline once, target it to collectors by attribute matchers (for example env="production" or a regex on region), and push it remotely over OpAMP so it hot-applies without restarting the collectors.

The skill provides end-to-end workflows built around the Fleet Management gRPC-Web API. It shows authoring a pipeline to a local file, validating it before shipping with `alloy fmt` and `alloy validate`, then creating it via the PipelineService API with attribute matchers, and verifying rollout by listing collectors and confirming each reports REMOTE_CONFIG_STATUS_APPLIED. A troubleshooting workflow surfaces collectors in REMOTE_CONFIG_STATUS_FAILED with their error messages, re-validates the offending pipeline, inspects the local Alloy UI on port 12345 for per-component health, and reads collector logs. An onboarding workflow documents the minimal local `remotecfg` bootstrap block (standalone and Helm), which points at the Fleet Management host with attributes and a poll frequency. API keys are supplied via bearer tokens and environment variables, and the collector attribute API lets matchers target env, team, and region labels.

Target users are platform and observability engineers operating many Alloy collectors who need centralized, hot-applied configuration, safe pre-deploy validation, and rollout verification across a large fleet. The operations are legitimate infrastructure administration using official Grafana tooling and standard token-based authentication.