Getting Started with Kollect

Kollect is a multi-cloud and infrastructure discovery tool that inventories resources across public clouds (AWS, Azure, GCP), cloud-native platforms (Kubernetes, OpenShift, Docker), virtualization (VMware vSphere, Hyper-V, Proxmox, Nutanix), and DevOps tooling (Terraform, Vault) — all from a single self-contained binary.

Download the binary

Grab the latest release for your platform. The binary is self-contained — there is nothing else to install.

# amd64 (use kollect-linux-arm64 for ARM, kollect-linux-386 for 32-bit)
curl -L -o kollect https://github.com/MichaelCade/kollect-web/releases/latest/download/kollect-linux-amd64
chmod +x kollect

# Apple Silicon (M1/M2/M3). For Intel Macs use kollect-darwin-amd64
curl -L -o kollect https://github.com/MichaelCade/kollect-web/releases/latest/download/kollect-darwin-arm64
chmod +x kollect

# If macOS quarantines the download:
xattr -d com.apple.quarantine kollect 2>/dev/null || true

# amd64 (use kollect-windows-arm64.exe for ARM)
Invoke-WebRequest -Uri "https://github.com/MichaelCade/kollect-web/releases/latest/download/kollect-windows-amd64.exe" -OutFile "kollect.exe"

# Multi-arch image (amd64 and arm64)
docker run --rm -p 8080:8080 ghcr.io/michaelcade/kollect:latest

Quick start

1. Launch the web interface

The fastest way to see Kollect is the browser UI. It starts a local server on http://localhost:8080 and opens your browser:

./kollect --browser

From there you can click a platform to connect to it, or use the import button to load a previously exported JSON file.

2. Collect a specific platform

Point Kollect at a single platform, or collect everything it can reach:

# One platform
./kollect --inventory aws

# Everything with detected credentials, in the browser
./kollect --inventory all --browser

# Valid inventory types:
# kubernetes, openshift, aws, azure, gcp, terraform, vault,
# docker, vsphere, proxmox, hyperv, nutanix, all

3. Explore snapshots, costs and AI

# Hunt snapshots across Kubernetes, AWS, Azure and GCP
./kollect --snapshots

# Start the MCP server for AI tools like Claude
./kollect --mcp

Command-line flags

Common flags are summarised below. See the Configuration guide for the complete list including every platform-specific flag and environment variable.

• --browser — open the web interface (can be used alone to import data)
• --inventory <type> — platform to collect (see list above)
• --snapshots — cross-platform snapshot hunter
• --mcp — start the MCP (Model Context Protocol) server
• --output <file> — save collected data to a JSON file
• --port <n> — web/API server port (default 8080; also KOLLECT_PORT)
• --scenario-csv <path> — export a Veeam Scenario Builder CSV
• --help — show all flags

What Kollect discovers

• AWS — EC2, S3, RDS, DynamoDB, VPCs, EBS snapshots
• Azure — VMs, Storage Accounts, VNets, SQL Databases, File Shares, CosmosDB
• GCP — Compute Instances, Storage Buckets, Cloud SQL, VPCs, disk snapshots
• Kubernetes / OpenShift — nodes, namespaces, pods, deployments, services, PVs, KubeVirt VMs
• Docker — containers, images, volumes, networks
• VMware vSphere — VMs, ESXi hosts, datastores, clusters, networks, resource pools, live performance metrics
• Hyper-V / Proxmox / Nutanix — virtual machines, hosts/nodes, storage and cluster inventory
• Terraform — resources, outputs and variables from local or remote state
• Vault — secrets engines, auth methods and policies

Next steps

1. Configure credentials and platform flags
2. Explore the local REST API
3. Troubleshoot connection issues

For help, visit the GitHub issues page.