- Home
- Skills
- DevOps & Cloud
- Infra Drift & Cost Sentinel
More screenshots
Works with the AI Agents you already use
Infra Drift & Cost Sentinel
Catches security drift, cost leaks, and state drift in your Terraform before you apply — with a prioritized fix list, not just a wall of warnings.
$12
Infra Drift & Cost Sentinel
Example session with this skill installed
Here's my Terraform for our app's infra — can you check it for security issues, wasted spend, or anything that looks like it drifted from what's actually deployed before I run apply?
- Read your context and instructions
- Compiled the infra drift
Stack summary: AWS Terraform project, 6 resources reviewed (S3, security group, EC2, EBS, RDS).
| Severity | Category | Resource | Issue | Fix |
|---|---|---|---|---|
| Critical | Security | aws_db_instance.main | publicly_accessible = true on a prod-named DB | Set publicly_accessible = false; access via VPC/bastion or SSM tunnel instead |
| Critical | Security | aws_s3_bucket_acl.logs_acl | acl = "public-read", no public-access-block resource present | Remove the ACL, add aws_s3_bucket_public_access_block with all four flags true |
| Critical | Security | aws_security_group.db_sg | Ingress on port 5432 open to 0.0.0.0/0 | Restrict cidr_blocks to the app tier's SG or VPC CIDR |
| High | Security | aws_db_instance.main | storage_encrypted = false | Set storage_encrypted = true (note: requires recreation — plan for a maintenance window) |
| High | Cost | aws_instance.web_dev | m5.4xlarge on a resource tagged Environment = dev | Right-size to t3.large/m5.large for dev workloads pending actual usage data |
| High | Cost | aws_ebs_volume.orphan | Volume defined with no aws_volume_attachment anywhere in the project | Confirm it's unused, then remove — orphaned EBS volumes bill indefinitely |
| Low | Hygiene | All resources | No tags/cost-allocation tags except web_dev | Add a consistent tagging block (Environment, Owner, CostCenter) so future cost tooling can attribute spend |
Top 3 actions:
- Lock down the DB — it's both public and unencrypted, the worst combination in this file.
- Close the security group's 5432 ingress to 0.0.0.0/0 before this ever reaches prod.
- Remove the public ACL on the logs bucket.
What wasn't checked: No terraform plan or state file was provided, so state-drift checks (manual changes, unexpected replace/destroy) couldn't run — only code-level checks. Sending a terraform plan -out=plan.tfplan output would sharpen that section.
Connects securely to your tools. The creator never sees your data.
About this skill
Most linters tell you what's wrong. This skill tells you what actually matters and in what order.
Feed it your Terraform files, a terraform plan output, or a CloudFormation/Pulumi template. It runs three passes — security drift (public buckets, open security groups, unencrypted storage, overly broad IAM), cost leaks (oversized dev instances, orphaned volumes, missing lifecycle policies), and state drift (unexpected replace/destroy, manual console changes not reflected in code) — then hands back a severity-ranked findings table with a real one-line fix for each issue, plus a "top 3 actions" callout.
No generic advice, no padded findings to look thorough — if something's fine, it says so. Built for AWS-first stacks with GCP/Azure pattern coverage included.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for DevOps & Cloud.
- No live cloud access: this skill only reasons over what you paste or upload (files, plan output, or state). It can't independently verify actual account configuration, so it will miss anything not represented in the input you give it.
- State-drift detection needs a terraform plan or state output to be reliable. Given only .tf files with no plan, drift findings are limited to what's inferable from code alone — manual console changes won't be caught.
- AWS has the deepest, most tested check coverage. GCP and Azure checks cover the same three categories (security, cost, drift) but with a narrower pattern set — treat those results as a solid first pass, not exhaustive.
- This is a judgment/triage layer, not a replacement for exhaustive static analysis tools (tflint, checkov, tfsec). It won't catch every syntax-level or provider-schema issue those tools specialize in — it's built to prioritize what matters, not enumerate everything.
- Cost findings are directional (flags mismatches and waste patterns), not precise billing estimates — it doesn't calculate actual dollar amounts or check current provider pricing.
- CloudFormation and Pulumi inputs are supported via the same check logic, but pattern coverage was built and tested primarily against Terraform syntax — expect Terraform to give the most accurate results.
- Very large projects (50+ resources) are summarized by module/count in the findings rather than listing every resource individually, to keep the report readable.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 2 months ago
Permissions
No special permissions declared or detected