Alloy

What it does

When your agent uses it

• Writing a config.alloy
• Replacing Grafana Agent / OTel Collector
• Scraping K8s pods
• Debugging Alloy isnt sending anything — even when the user says set up the agent

Example prompts

• “Alloy isn”
• “— even when the user says”
• “write me a scrape config”
• “/alloy”

Requirements

• Docker
• A credential in GRAFANA_API_KEY

Workflow steps

3 steps, taken from the step headings in SKILL.md.

1. 1Write + validate a config locally
2. 2Add log shipping (file → Loki)
3. 3Receive OTLP traces and ship to Tempo

What it can do on your machine

Read from SKILL.md and the folder at commit 1ccacf2. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  Shell commands in SKILL.md call:

  • curl
  • jq
  • brew
  • apt

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  Links to these hosts (documentation or services it may open):

  • grafana.com

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names these keys or tokens, usually read from environment variables:

  • GRAFANA_API_KEY

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Alloy loads about 1.3k tokens when it runs, and up to ~11k if it reads all its reference files. Until then it costs about 238 tokens; SKILL.md has 164 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~238

When it runs · the whole SKILL.md, loaded when a task matches

~1.3k

With references · SKILL.md plus every file in references/, read only if the agent opens them

~11k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from grafana/skills at commit 1ccacf2, republished under its Apache-2.0 licence (© grafana). 164 words, ~1,321 tokens.

name

alloy

description

Build a unified telemetry pipeline with Grafana Alloy — one OpenTelemetry-compatible binary that collects metrics, logs, traces, and profiles and ships to Grafana Cloud / Prometheus / Loki / Tempo / Pyroscope. Covers the Alloy config language (blocks, `sys.env`, component refs), `prometheus.scrape` → `remote_write`, `loki.source.file` + `loki.process` → `loki.write`, `otelcol.receiver.otlp` → `otelcol.exporter.otlp`, `pyroscope.scrape`, K8s / Docker / EC2 discovery, relabeling, modules (`import.file/git/http`), clustering, Fleet Management `remotecfg`, the Alloy UI at `:12345`, and `alloy fmt` / `alloy validate`. Use when writing a `config.alloy`, replacing Grafana Agent / OTel Collector, scraping K8s pods, parsing logs, ingesting OTLP, or debugging "Alloy isn't sending anything" — even when the user says "set up the agent", "write me a scrape config", "drop these logs before sending", or "OTel collector config" without naming Alloy.

license

Apache-2.0

// config.alloy — metrics → Grafana Cloud
prometheus.scrape "app" {
  targets = [{"__address__" = "localhost:9090"}]
  forward_to = [prometheus.remote_write.cloud.receiver]
  scrape_interval = "30s"
}

prometheus.remote_write "cloud" {
  endpoint {
    url = sys.env("PROMETHEUS_URL")
    basic_auth {
      username = sys.env("PROM_USER")
      password = sys.env("GRAFANA_API_KEY")
    }
  }
}

# 1. Format + syntax-check (catches typos before run)
alloy fmt config.alloy
alloy validate config.alloy

# 2. Run it
alloy run config.alloy
# or as a service: systemctl restart alloy

# 3. Verify all components are healthy via the UI on port 12345
curl -s http://localhost:12345/api/v0/web/components \
  | jq '.[] | select(.health.state != "healthy") | {id, state:.health.state, msg:.health.message}'
# Expect: empty (nothing unhealthy). Otherwise the row shows the failing component + reason.

# 4. Verify samples are flowing
curl -s http://localhost:12345/metrics \
  | grep -E '^prometheus_remote_storage_(samples_total|enqueue_retries_total)' | head
# samples_total should be > 0 and rising; retries should be 0.

loki.source.file "app_logs" {
  targets    = [{ __path__ = "/var/log/app/*.log", job = "app" }]
  forward_to = [loki.write.cloud.receiver]
}

loki.write "cloud" {
  endpoint {
    url = sys.env("LOKI_URL")
    basic_auth {
      username = sys.env("LOKI_USER")
      password = sys.env("GRAFANA_API_KEY")
    }
  }
}

# Verify in Grafana → Explore → Loki:
#   {job="app"}
# Expect lines streaming. If empty:
#   - check /var/log/app/*.log actually exists + readable by the alloy user
#   - http://localhost:12345 → loki.source.file.app_logs → "Targets" tab

otelcol.receiver.otlp "default" {
  grpc { endpoint = "0.0.0.0:4317" }
  http { endpoint = "0.0.0.0:4318" }
  output { traces = [otelcol.exporter.otlp.tempo.input] }
}

otelcol.exporter.otlp "tempo" {
  client {
    endpoint = "tempo-xxx.grafana.net/tempo:443"
    auth     = otelcol.auth.basic.grafana_cloud.handler
  }
}

otelcol.auth.basic "grafana_cloud" {
  username = sys.env("TEMPO_USER")
  password = sys.env("GRAFANA_API_KEY")
}

# Verify in Grafana → Explore → Tempo → service.name = your-service
# At the Alloy level, watch the receiver:
curl -s http://localhost:12345/metrics | grep otelcol_receiver_accepted_spans

© grafana, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 3 other files (references) in skills/grafana-core/alloy of grafana/skills.

• SKILL.md
• references/collection-patterns.md
• references/components.md
• references/config-syntax.md

Open the folder on GitHubat commit 1ccacf2