dotfiles

Analyze the resource usage of a Kubernetes Deployment, StatefulSet, or DaemonSet over the last 30 days and recommend right-sized CPU/memory requests and limits plus autoscaling (KEDA/HPA) improvements. Triggers when the user asks to "right-size", "analyze resource usage", "tune requests/limits", "is this workload over/under-provisioned", or "improve autoscaling" for a named workload. Reads live cluster spec and 30-day metrics through a single Grafana instance (delegating to the sre-grafana and sre-kubernetes skills), treats the live cluster as authoritative and an optional manifest as the verification + apply target, and never edits cluster resources directly. Detects common database and JVM engines (MongoDB, PostgreSQL, Redis, Kafka, ClickHouse, Elasticsearch/OpenSearch, generic JVM) and sizes memory from the engine cache/heap model instead of raw working set. Also recognises page-cache-sensitive services (Vault, etcd, Loki) that thrash without OOM/PSI, using major page faults as the memory guard.

Write and maintain Architecture Decision Records (ADRs) following best practices for technical decision documentation. Use when documenting significant technical decisions, reviewing past architectural choices, or establishing decision processes.

This skill should be used BEFORE running any git commit command. Triggers when about to run `git commit`. Ensures commit messages follow Conventional Commits specification.

Use when addressing the review feedback left on a GitHub pull request - reads all review comments, suggestions, and conversation, validates each one, and implements the valid ones as local edits after per-item confirmation. Triggers on "address PR review comments", "implement review feedback", "apply PR suggestions", "go through / resolve the comments on my PR", or "fix review comments". This skill CONSUMES and IMPLEMENTS reviewer feedback - it does NOT author review comments (that is the github-pr-review skill).

Use when reviewing GitHub pull requests with gh CLI - creates pending reviews with code suggestions, batches comments, and chooses appropriate event types (COMMENT/APPROVE/REQUEST_CHANGES)

Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".

This skill should be used when the user wants to write, create, or generate a Jira ticket. It produces well-structured tickets with a user story, background paragraph, and acceptance criteria checklist — shown as Markdown and created in Jira via Atlassian Document Format (ADF) so they render cleanly in the web UI.

Analyze an SRE alert (typically from Grafana Alertmanager) end-to-end: explain why it is firing, correlate evidence from metrics, logs, traces, profiles and the Kubernetes cluster, identify the most likely root cause, and propose a concrete fix. Triggers when the user pastes an alert payload, an alert URL, or asks to "investigate", "analyze", "debug", or "find the root cause" of an alert. Delegates data collection to the `sre-grafana` and `sre-kubernetes` skills.

Create, modify, and organise Grafana dashboards including panels, variables, transformations, and annotations. Dashboards are authored as local JSON files by default and only pushed to Grafana when explicitly requested. Use when the user asks to create a Grafana dashboard, add a panel, configure a time series or stat panel, add template variables, set up dashboard linking, use transformations, configure thresholds, build a dashboard for a service, add an annotation to a dashboard, or export dashboard JSON. Triggers on phrases like "create dashboard", "add panel", "time series panel", "Grafana dashboard JSON", "template variables", "dashboard variable", "panel transformation", "threshold", "stat panel", "table panel", "Grafana annotations", or "dashboard folder".

Query a Grafana instance for metrics, logs, traces, and profiles to investigate an alert or answer related questions. Triggers when the user asks to "check Grafana" or "query metrics/logs/traces/profiles", references a Prometheus, VictoriaMetrics, VictoriaLogs, VictoriaTraces, Tempo, or Pyroscope datasource exposed through Grafana, or when the `sre-analyze-alert` skill needs supporting evidence. Selects the appropriate datasource, runs the query against Grafana's HTTP API, and summarizes the result.

Inspect a Kubernetes cluster — reachable via an API server URL and bearer token, a context name from `~/.kube/config`, or a Grafana datasource proxy — for failing pods, warning events, recent rollouts, resource pressure, and other state relevant to an SRE alert or question. Triggers when the user asks about pod / deployment / node health, references a cluster URL + token, names a kubeconfig context, points at a Grafana instance with a Kubernetes datasource, or when `sre-analyze-alert` needs cluster-side evidence. Read-only by default; destructive actions require explicit confirmation.