← ClaudeAtlas

cloud-resource-healthlisted

Use when diagnosing an unhealthy or degraded cloud resource from provider status, metrics, logs, dependencies, deployments, and permissions. Start with bounded read-only evidence and produce a classified remediation plan.
sandbaseai/workbuddy-skill · ★ 9 · AI & Automation · score 80
Install: claude install-skill sandbaseai/workbuddy-skill
# Cloud Resource Health ## Purpose and safety boundary Diagnose a named cloud resource using evidence from the resource, its dependencies, recent changes, and the affected user path. This Skill is read-only by default. It does not reboot, resize, delete, change IAM/network policy, alter concurrency, or enable logging. Any remediation must have an owner, authorization, impact window, rollback/recovery plan, and post-change verification. Do not assume that a provider status of “available” means the workload is healthy. Resolve the exact account/project, region, resource identifier, tenant scope, and time window before drawing conclusions. Redact account identifiers, customer data, tokens, private addresses, and raw log payloads. ## Inputs Capture: - exact error, affected user journey, first/last occurrence, frequency, and severity; - provider, account/project, region, resource type and identifier, environment, and owner; - recent deploys, config/IAM/network changes, scaling events, provider events, and incident timeline; - relevant health metrics, logs, traces, dependency status, quotas, limits, and alert state; - baseline comparison window and any concurrent changes or known data gaps. If multiple resources match, stop and request a narrower scope. If telemetry is missing, report the blind spot and lower confidence rather than treating “no data” as “healthy.” ## Workflow 1. **Preserve and scope the symptom.** Create a short sanitized timeline and identify the failing