Env Diff
Find out why it works in staging but not production
Compares configurations across your environments — Terraform state, Helm values, env vars, K8s resources — and surfaces every difference. The fastest way to find the manual change someone made in production 6 months ago.
INGREDIENTS
PROMPT
Create a skill called "Env Diff". Compare configurations across environments to find drift: 1. Compare Terraform variables (tfvars) across workspaces 2. Diff Helm values.yaml files across environments 3. Compare K8s resources using `kubectl diff` or by fetching and comparing manifests 4. Diff environment variables from deployment configs 5. Identify resources that exist in one environment but not another For each difference found: - Classify as: intentional (different sizing), drift (manual change), or missing (exists in one env only) - Flag security-relevant differences (IAM, security groups, encryption, network policies) - Suggest whether to align or document the difference Present as a clean diff report, organized by severity.
How It Works
Environment drift causes outages that are uniquely frustrating because the
code is the same — it's the configuration that differs. This skill diffs
your environments systematically.
What You Get
- Side-by-side comparison of Terraform state/vars across workspaces
- Helm values.yaml diff across environments (dev, staging, prod)
- Kubernetes resource comparison (deployments, configmaps, secrets, ingresses)
- Environment variable comparison across deployment configs
- Identification of resources that exist in one environment but not another
- Flagging of security-relevant differences (open ports, permissive IAM, missing encryption)
Setup Steps
- Tell your Claw which environments to compare (e.g., staging vs production)
- Point it at the relevant configs, Terraform workspaces, or K8s clusters
- Review the diff report
- Decide for each difference: align environments or document why they differ
Tips
- Pay special attention to security-group and IAM differences — these are the dangerous ones
- Some differences are intentional (smaller instances in dev) — mark these as acknowledged
- Run before deployments to catch drift that could cause issues
- Pairs well with Drift Hunter for Terraform-specific drift detection