How CloudSlasher Works: 5-Stage Deterministic Safety Pipeline
- Non-Invasive Tenancy Ingestion: Zero-agent, read-only IAM scan maps running topology across compute instances, storage volumes, databases, and network gateways.
- FOCUS 1.0 Lakehouse Analytics: Normalizes billing records with sub-25ms vectorized DuckDB queries to identify idle and oversized resources.
- Interactive What-If Modeling: Compound multi-vector cost simulator reveals immediate and 12-month runway savings.
- Targeted Terraform Synthesizer: Generates declarative, reviewable HCL patches with automatic companion rollback modules.
- 1-Click GitOps Pull Request: Dispatches non-destructive PRs to your GitHub repository with zero direct-write access to production main branches.
Cloud & Infrastructure Support Matrix
| Cloud Platform | Supported Resource Types | Detection Scope | Remediation Mechanism | Auto-Delete Risk |
|---|---|---|---|---|
| Oracle Cloud (OCI) | Compute VMs, Autonomous DB (ATP/ADW), Block Volumes, NAT Gateways, Ephemeral IPs | Idle shapes, Always Free Ampere A1.Flex migration, DB ECPU conversion, Lower Cost VPU tiering | Tested Terraform HCL + GitOps Pull Request | 0% Risk (Strictly Read-Only) |
| Amazon Web Services (AWS) | EC2, EBS Volumes, RDS, Application Load Balancers, NAT Gateways, Elastic IPs | Overprovisioned instances, unattached gp2/gp3 volumes, idle ALBs, public IPv4 hourly charges ($0.005/hr), idle NAT Gateways ($33/mo/AZ) and missing S3 VPC endpoints | OpenTofu / Terraform Remediation PR | 0% Risk (Strictly Read-Only - Never Shuts Down Without PR) |
| Microsoft Azure | Azure VMs, Managed Disks, Azure SQL, Public IP addresses | Unused standard disks, idle VM rightsizing | Declarative HCL Remediation Patch | 0% Risk (Strictly Read-Only) |
| Google Cloud (GCP) | Compute Engine, Persistent Disks, Cloud SQL | Zonal SSD rightsizing, idle instances | Terraform GitOps PR | 0% Risk (Strictly Read-Only) |
| Kubernetes (EKS/AKS/GKE/OKE) | Worker nodes, pod resource limits vs requests, PV claims | Oversized node pools, runaway container allocations | K8s manifest / Helm HCL patch | 0% Risk (Strictly Read-Only) |
How CloudSlasher Compares with Infracost, Harness, Finout, Holori, and CAST AI
- CloudSlasher vs. Infracost: Infracost operates shift-left inside CI/CD to estimate proposed cost diffs on pull requests. CloudSlasher operates post-deployment on live fleets across OCI, AWS, Azure, GCP, and Kubernetes to discover active waste that was already provisioned.
- CloudSlasher vs. Harness & Finout: Harness and Finout are cost allocation and dashboarding tools that show where money went after the fact. CloudSlasher actually identifies specific actionable waste (idle compute, orphaned disks, unattached IPv4s, overprovisioned DB shapes) and writes the exact Terraform fix.
- CloudSlasher vs. CAST AI: CAST AI uses automated agents to resize and shut down Kubernetes nodes on the fly, introducing stability risks for stateful or latency-sensitive workloads. CloudSlasher is 100% read-only across all clouds and never shuts down live infrastructure autonomously; all remediation is submitted as reviewable GitOps pull requests.
- CloudSlasher vs. Holori: Holori focuses on visual cloud architecture modeling. CloudSlasher provides live telemetry-backed cost simulation, automated CIS compliance auditing, and executable Infrastructure-as-Code remediation.