SRE-Skills
AGENTS.md4 starsChanged 23 days agoMITView on GitHub
This repository is a generic SRE skill pack. Each skill is a folder with a SKILL.md (Agent Skills format). Load the matching skill before acting. Never run a write command, call an MCP write tool, or post an external message without explicit human approval in this conversation. Skills marked "manual only" must be started by the user (for example /sre-chaos-engineering, or /sre:sre-chaos-engineering in a Claude Code plugin install); never run their steps on your own. Script and reference paths inside…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 32,671 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository84 other files it gives its agents.
- .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-prometheus-grafanaSkill
- 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 AGENTS.md into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Diagnose read-only first: Use only read commands and read-only MCP tools (get, describe, logs, history, plan, queries, dashboards).
- Very specific: line 11
Act: Run exactly what was approved — nothing more.
- Distinctive: line 13
Stop and escalate if worse: If metrics degrade, the action fails, or the outcome is unexpected: stop, do not try further fixes on your own, …
≈ “Common” is an estimate: at least 1 in 5 of the 41,581 AGENTS.md 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
<!-- GENERATED by scripts/build-targets.py from .claude/skills/*/SKILL.md. Do not edit; edit the skills and re-run. -->SRE skill pack (AGENTS.md: Codex, Cursor, Grok, Copilot)
This repository is a generic SRE skill pack. Each skill is a folder with a SKILL.md (Agent Skills format). Load the matching skill before acting.
Safety gates (always apply, for every skill and every tool)
- Diagnose read-only first. Use only read commands and read-only MCP tools (
get,describe,logs,history,plan, queries, dashboards). Change nothing. - Propose with risk and reversibility. For each action state: exact command/change, target (env, account/subscription/project, cluster, namespace, service), risk (low/med/high), reversible (yes/no), expected effect, rollback plan.
- Wait for explicit human approval. Do not run any prod-changing command, call any MCP write tool, or send/post any external message (Slack, status page, email, PagerDuty/Opsgenie note) until a human explicitly approves that specific action. Silence, "looks fine", or approval of a different action is not approval.
- Act. Run exactly what was approved — nothing more. Prefer dry-run/diff/plan output first.
- Verify. Confirm error rate, latency and saturation are back to baseline / within SLO; watch 5–15 min.
- Stop and escalate if worse. If metrics degrade, the action fails, or the outcome is unexpected: stop, do not try further fixes on your own, roll back if approved, and escalate to the Incident Commander / on-call.
Never run a write command, call an MCP write tool, or post an external message without explicit human approval in this conversation. Skills marked "manual only" must be started by the user (for example /sre-chaos-engineering, or /sre:sre-chaos-engineering in a Claude Code plugin install); never run their steps on your own.
Script and reference paths inside a skill (for example scripts/az-ro.sh or ../sre-k8s-triage/scripts/...) are relative to that skill's folder, the folder containing its SKILL.md. Resolve them against that folder: cd into it, or prefix the skill path. Older copies may write ${CLAUDE_SKILL_DIR}, which means the same folder; substitute it yourself if your tool doesn't.
Skills live in .claude/skills/ (read natively by Claude Code, Cursor, Copilot and Grok). Codex reads .agents/skills/: install per docs/install/codex.md.
Skills
| Skill | Use when | Invocation |
sre-aws | Use when the system runs on AWS: Health, CloudWatch metrics and Logs Insights, EC2/EKS/RDS/ELB/ASG/ECS status, CloudTrail, quotas. Read-only profile via AWS CLI or MCP; writes need approval. | auto |
sre-azure | Use when the system runs on Azure: Resource Health, Azure Monitor, Log Analytics (KQL), AKS/VM/App Service status, Activity Log. Read-only roles via az CLI or Azure MCP; writes need approval. | auto |
sre-capacity-planning | Use before launches or peak events, or when saturation trends up: forecast demand from metrics, compute headroom, plan load tests, set scaling triggers. Read-only queries. | auto |
sre-change-safety | Use before, during or right after a deploy, migration or infra change: risk score, freeze and error-budget check, canary/progressive rollout, rollback plan, post-deploy verification. | auto |
sre-chaos-engineering | Use to design or run a chaos / fault-injection experiment or game day, with hard no-go rules, approval and abort criteria. Manual invocation only. Not for live incident response. | manual only |
sre-cloud-cost-review | Use for cloud or observability cost reviews and spend spikes: idle/oversized resources, storage sprawl, data transfer, commitments, ingest/retention. Read-only; never deletes or resizes without approval. | auto |
sre-database-triage | Use when Postgres, MySQL/MariaDB or Redis is slow or failing: connection exhaustion, locks, long or slow queries, replication lag, disk/WAL/bloat, Redis memory/eviction, DB CPU. Read-only; kill and failover are gated. | auto |
sre-datadog | Use when metrics, traces, logs, monitors or SLOs live in Datadog: queries, monitor and SLO review, incidents and events. Read-only MCP or API (read-scoped keys); writes need approval. | auto |
sre-dependency-mapping | Use for PRRs, reliability reviews, incident blast radius or chaos planning: map upstream/downstream services, criticality, timeouts/retries and single points of failure. Read-only. | auto |
sre-diagnostics | Use to investigate a live issue read-only: logs/metrics/traces queries, correlation, hypothesis testing, profiling and bottleneck identification. Gives exact commands when tools are unavailable. | auto |
sre-dr-backup-test | Use when testing backups or DR readiness: backup-restore and failover drills against RTO/RPO, restoring only to new isolated targets after approval, with an evidence record. | auto |
sre-dynatrace | Use when data lives in Dynatrace: DQL for problems, logs, metrics and entities, Davis problem triage, SLO and health checks. Read-only MCP or DQL for Notebooks. | auto |
sre-emudhra-ca-check | Use to check whether a TLS/code-signing cert chains through one of the 8 eMudhra intermediate CAs revoked for private-key compromise, Oct 2026 (Mozilla bug 2083449): endpoints, cert files, k8s TLS secrets. Read-only; a match needs a gated reissue. | auto |
sre-engineer | Use for live problems (incident, alert, latency or 5xx, gated remediation, runbook walk) and for SRE design: observability architecture, reliability reviews, on-call program design, and choosing tools. | auto |
sre-expiry-audit | Use when asked what expires soon: TLS certs (endpoints, cert-manager), domain registrations, ACM/IAM certs, Secrets Manager rotation, GCP Secret/Certificate Manager, Azure Key Vault. Read-only metadata; flags 30/14/7 days. | auto |
sre-gcp | Use when the system runs on Google Cloud: Cloud Monitoring, Cloud Logging, GKE/Compute/Cloud Run/Cloud SQL status, audit logs, quotas. Viewer roles via gcloud or MCP allowlist; writes need approval. | auto |
sre-incident-command | Use when an incident is declared or suspected: declare, assign roles (IC, ops, comms, scribe), set severity, keep the timeline, drive decisions, hand off and close. Debugging goes to sre-engineer. | auto |
sre-incident-communication | Use when an incident update or stakeholder message is needed: announcements, status updates, resolution notices, executive summaries with severity cadence. Drafts only until approved. | auto |
sre-k8s-safe-remediation | Use when a fix from sre-k8s-triage needs a kubectl or helm write (rollback, restart, scale, pod delete, Helm rollback): evidence, GitOps/HPA ownership check, preview, explicit approval, verify. | manual only |
sre-k8s-triage | Use when a Kubernetes workload or node is unhealthy: CrashLoopBackOff, Pending, OOMKilled, ImagePullBackOff, node pressure, DNS, probes, stuck rollouts, Helm, and whether a restart is safe. Read-only decision tree. | auto |
sre-k8s-upgrade-autopilot | Plan a Kubernetes minor-version upgrade - scripted preflight, support-status check, go/no-go, gated EKS/GKE/AKS steps, verification after every step. Scripts tested on kind. Use for "upgrade our cluster to 1.xx". | auto |
sre-message-queue-triage | Use when Kafka, RabbitMQ or SQS/SNS is backing up or dropping messages: consumer lag, DLQ growth, broker disk/under-replication, redelivery storms, poison-pill messages. Read-only; purge, rebalance and resize are gated. | auto |
sre-migration-safety | Review a SQL schema migration for PostgreSQL or MySQL/MariaDB before it merges. Says which statements lock, rewrite or block writes, and gives the safe pattern for each. Never connects to production. | auto |
sre-observability-cost-cutter | Use to reduce observability spend. Read-only usage and cost audit (works with Datadog), a ranked savings report with approval-gated changes and rollbacks, and conversion of metric monitors to Prometheus rules checked by promtool. | auto |
sre-observability-query-cookbook | Use when you need a correct query for golden signals, RED/USE, saturation, error ratio, latency percentiles or SLO burn rate in PromQL, DQL, Datadog or LogQL. | auto |
sre-oncall-audit | Use before rotation changes, after a migration, or quarterly: read-only audit of PagerDuty/Opsgenie schedules and escalation policies for coverage gaps, single-person layers, DST, loops, unreachable users. | auto |
sre-oncall-handoff | Use at shift change or when asked for an on-call summary: open incidents, alerts that fired, noisy alerts, ongoing risks, pending changes and follow-ups. Read-only. | auto |
sre-pagerduty-opsgenie | Use for on-call, incident and handoff tasks in PagerDuty or Opsgenie: who is on call, open incidents/alerts, escalation policies, timelines, capped page counts. Writes need approval. | auto |
sre-php2shell-check | Use to check PHP hosts/containers for PHP2Shell (CVE-2026-109618, CVE-2026-109619), the Oct 2026 critical unauthenticated RCE pair, CVSS 9.2. Read-only version check (php -v, package manager, X-Powered-By); a hit needs a gated upgrade. | auto |
sre-postmortem-templates | Use after an incident is resolved to draft or facilitate a blameless postmortem / incident review / RCA with a timeline from paging, monitoring and deploy history. Not for active incidents. | auto |
sre-production-readiness | Use before a new service or major change goes live: run a production readiness review (PRR) / launch checklist and produce a pass/fail verdict with evidence. | auto |
sre-prometheus-grafana | 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. | auto |
sre-remediation-actions | Reference of gated recovery procedures (rollback, scale, restart, failover) with risk/reversibility labels, evidence capture and dry-run first. Loaded by sre-engineer before any remediation. | reference (loaded by other skills) |
sre-runbook-templates | Use when creating or reviewing a runbook: full template with owner, scope, diagnosis, decision points, gated actions with rollback, verification, escalation and last-reviewed date. | auto |
sre-safety-gates | Shared approval gates for every sre-* skill: read-only first, propose with risk/reversibility, explicit human approval before any prod change, MCP write or external post, verify, stop if worse. | reference (loaded by other skills) |
sre-security-incident-triage | Use when a security compromise is suspected: leaked credentials, suspicious access, compromised host or container, unexpected infra changes. Preserve evidence, gated containment, hand off to security. | auto |
sre-service-mesh-triage | Use when Istio or Linkerd sidecars are suspected: 503s from the mesh vs. the app, mTLS handshake failures, retry/circuit-breaker storms, sidecar injection failures, egress blocked by policy. Read-only; fixes hand off to sre-k8s-safe-remediation. | auto |
sre-slo-alerting | Use when creating or revising SLIs, SLOs, error budget policy, burn-rate alerts or paging policy, and for alert-fatigue reduction. Not for responding to a firing alert (use sre-engineer). | auto |
sre-slo-burn-rate-rules | Use when authoring or fixing Prometheus SLO burn-rate alert rules: generate 4-tier multiwindow rules from an SLO target and counters, then promtool lint and test. Files only, never deploys. Not for responding to a firing alert (use sre-engineer). | auto |
sre-terraform | Use for any Terraform change, plan review, drift check or rollback: confirm target, fmt/validate, saved-plan review, state inspection without leaking secrets, gated apply of the reviewed plan. | auto |
sre-toil-review | Use for quarterly ops reviews or when on-call load is high: inventory repetitive manual work, score frequency x time x risk, and pick automation candidates. | auto |
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.

