sre-prometheus-grafana
Skill4 starsChanged 25 days agoMITView on GitHub
Use when metrics, logs or alerts live in Prometheus, Alertmanager, Loki or Grafana: bounded PromQL/LogQL, active alerts, rule linting with promtool, gated silences, dashboards.
- Reads credentials
In this repository84 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .github/copilot-instructions.mdCopilot instructions
- .cursor/rules/sre-safety-gates.mdcCursor rule
- sre-awsSkill
- sre-azureSkill
- sre-capacity-planningSkill
- sre-change-safetySkill
- sre-chaos-engineeringSkill
- sre-cloud-cost-reviewSkill
- sre-database-triageSkill
- sre-datadogSkill
- sre-dependency-mappingSkill
- sre-diagnosticsSkill
- sre-dr-backup-testSkill
- sre-dynatraceSkill
- sre-emudhra-ca-checkSkill
- sre-engineerSkill
- sre-expiry-auditSkill
- sre-gcpSkill
- sre-incident-commandSkill
- sre-incident-communicationSkill
- sre-k8s-safe-remediationSkill
- sre-k8s-triageSkill
- sre-k8s-upgrade-autopilotSkill
- sre-message-queue-triageSkill
- sre-migration-safetySkill
- sre-observability-cost-cutterSkill
- sre-observability-query-cookbookSkill
- sre-oncall-auditSkill
- sre-oncall-handoffSkill
- sre-pagerduty-opsgenieSkill
- sre-php2shell-checkSkill
- sre-postmortem-templatesSkill
- sre-production-readinessSkill
- sre-remediation-actionsSkill
- sre-runbook-templatesSkill
- sre-safety-gatesSkill
- sre-security-incident-triageSkill
- sre-service-mesh-triageSkill
- sre-slo-alertingSkill
- sre-slo-burn-rate-rulesSkill
- sre-terraformSkill
- sre-toil-reviewSkill
- sre-awsSkill
- sre-azureSkill
- sre-capacity-planningSkill
- sre-change-safetySkill
- sre-chaos-engineeringSkill
- sre-cloud-cost-reviewSkill
- sre-database-triageSkill
- sre-datadogSkill
- sre-dependency-mappingSkill
- sre-diagnosticsSkill
- sre-dr-backup-testSkill
- sre-dynatraceSkill
- sre-emudhra-ca-checkSkill
- sre-engineerSkill
- sre-expiry-auditSkill
- sre-gcpSkill
- sre-incident-commandSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Wait for explicit human approval before any prod-changing command, MCP write tool, or external post.
- Very specific: line 112
Native histograms: if enabled, histogram_quantile(0.95, sum by (job) (rate(<metric>[5m]))) works without _bucket/le.
- Distinctive: line 124
If neither the Grafana MCP server nor the CLIs are available, output the queries for the human and stop.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Prometheus and Grafana
Safety gates (REQUIRED)
- Read-only diagnosis first.
- Propose each change with risk (low/med/high) and reversibility.
- Wait for explicit human approval before any prod-changing command, MCP write tool, or external post.
- Act only as approved, verify, and stop and escalate if anything gets worse.
Full rules: sre-safety-gates. Read-only MCP setup: sre-safety-gates/references/mcp.md.
Process skills: sre-slo-alerting, sre-engineer, sre-diagnostics. More queries: sre-observability-query-cookbook.
Rules
- Ad-hoc queries need a selector and a bounded range (e.g.
{job="<svc>"},[5m],--since=1h --limit=100). No bare metric names over long ranges. - Build long windows (1d, 3d) from recording rules, not raw ad-hoc queries.
- Never call
/-/reload,/-/quit, or the admin TSDB APIs (/api/v1/admin/tsdb/*: delete_series, clean_tombstones, snapshot). - Grafana service account tokens: Viewer (or custom read) role, with an expiry.
- Silences: preview with
amtool alert queryfirst; require--authorplus a ticket or approval; no empty or regex-only matchers (.*silences everything).
Access
- MCP: Grafana MCP server, pinned, with
--disable-write(seesre-safety-gates/references/mcp.md). - CLI fallback:
promtool,amtool,logcli, or the Prometheus HTTP API.
Read-only commands
# Instant query via HTTP API (selector required)
curl -sG "$PROM_URL/api/v1/query" --data-urlencode 'query=up{job="<svc>"} == 0'
# Active alerts (pending AND firing)
curl -s "$PROM_URL/api/v1/alerts" | jq -c '.data.alerts[] | {n:.labels.alertname,s:.state,j:.labels.job}' | head -50
# Firing only
curl -sG "$PROM_URL/api/v1/query" --data-urlencode 'query=ALERTS{alertstate="firing"}'
amtool alert query --alertmanager.url="$AM_URL" | head -50
amtool silence query --alertmanager.url="$AM_URL" | head -50
# Validate rule files before any change
promtool check rules --lint=all rules/*.yml
promtool test rules tests/*.yml
# Loki (bounded)
LOKI_ADDR=<loki-url> logcli query --since=1h --limit=100 '{app="<svc>"} |~ "(?i)error"'Write commands (approval required)
# 1. Preview exactly what the matchers hit
amtool alert query alertname="<name>" service="<svc>" --alertmanager.url="$AM_URL"
# 2. Silence with author + ticket, explicit matchers, short duration
amtool silence add alertname="<name>" service="<svc>" --author="<you>" --duration=1h --comment="<ticket/incident ID>" --alertmanager.url="$AM_URL"
amtool silence expire <silence-id> --alertmanager.url="$AM_URL"Multiwindow burn-rate rules (99.9% SLO)
Generate the rules with scripts/gen-burn-rate-rules.py rather than writing them by hand. The excerpt below is from its output, references/slo-burn-rate.rules.yml, which passes promtool check rules --lint=all and the unit tests in references/slo-burn-rate.test.yml. The recording rules are named slo:sli_error:ratio_rate<window> (level:metric:operations), aggregate sum by (job), and carry an slo label. The alerts are SLOErrorBudgetBurnFast (page) and SLOErrorBudgetBurnSlow (ticket). Thresholds are N * (1 - SLO); for 99.9%, 1 - 0.999 = 0.001.
| Tier | Burn | Long / short window | Severity |
| 1 | 14.4x | 1h / 5m | page |
| 2 | 6x | 6h / 30m | page |
| 3 | 3x | 1d / 2h | ticket |
| 4 | 1x | 3d / 6h | ticket |
python3 scripts/gen-burn-rate-rules.py --slo-name example-availability --objective 0.999 \
--errors 'http_requests_total{job="example-svc",code=~"5.."}' --total 'http_requests_total{job="example-svc"}' \
--out slo-example-availability.rules.yml - record: slo:sli_error:ratio_rate5m # also 30m, 1h, 2h, 6h, 1d, 3d
expr: |
sum by (job) (rate(http_requests_total{job="example-svc",code=~"5.."}[5m]))
/
sum by (job) (rate(http_requests_total{job="example-svc"}[5m]))
labels:
slo: "example-availability"
- alert: SLOErrorBudgetBurnFast # SLOErrorBudgetBurnSlow uses 1d/2h at 3x and 3d/6h at 1x
expr: |
(
slo:sli_error:ratio_rate1h{slo="example-availability"} > (14.4 * 0.001)
and
slo:sli_error:ratio_rate5m{slo="example-availability"} > (14.4 * 0.001)
)
or
(
slo:sli_error:ratio_rate6h{slo="example-availability"} > (6 * 0.001)
and
slo:sli_error:ratio_rate30m{slo="example-availability"} > (6 * 0.001)
)
labels:
severity: page
slo: "example-availability"The generated alerts have no for:, because the short window already confirms the burn is ongoing. Pass --for 2m only if you want a noise filter.
To check the current burn rate when an alert fires, query the same names: slo:sli_error:ratio_rate5m{slo="<slo>"} / (1 - <objective>).
Deploy through Git, not by hand; never /-/reload from this skill.
Version and deprecation notes
- Prometheus 3: range selectors are left-open; keep
rate()windows ≥ 4× the scrape interval, and use$__rate_intervalin Grafana panels. - Native histograms: if enabled,
histogram_quantile(0.95, sum by (job) (rate(<metric>[5m])))works without_bucket/le. - Promtail reached EOL on 2026-03-02; use Grafana Alloy.
- Grafana API keys were replaced by service account tokens;
GRAFANA_API_KEYis deprecated in mcp-grafana. - Legacy Grafana alerting was removed in Grafana v11 (use Grafana Alerting).
Bundled scripts and references
- Generate SLO burn-rate rules:
python3 scripts/gen-burn-rate-rules.py --help(or/sre-slo-burn-rate-rules). Self-test:PROMTOOL=<path> bash scripts/test-gen-burn-rate-rules.sh. - Syntax-check every PromQL code block in Markdown:
python3 scripts/validate-promql-snippets.py --promtool <path> <files>. - Tested example + unit test: references/slo-burn-rate.rules.yml, references/slo-burn-rate.test.yml
- references/promql-cheatsheet.md · references/logql-cheatsheet.md · references/safe-operations.md
If neither the Grafana MCP server nor the CLIs are available, output the queries for the human and stop.
- Queries/rules used (with selector and range), lint/test results, and any proposed rule or silence change with risk and approval status
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

