Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.murmur.dev/llms.txt

Use this file to discover all available pages before exploring further.

Environments and placements are the two catalog resources that control the compute substrate for agent VMs. An environment defines what a VM looks like. A placement defines where it runs.

​
Environments

An environment (environment kind) specifies the VM configuration:
FieldDescriptionExample
substrateCloud provider typeGCP, AWS
machine_type_refReference to a machine-type resourcen2-standard-8
disk_size_gbRoot disk size (0 = image default)100
cat <<EOF | murmur set environment large
name: large
substrate: GCP
machine_type_ref: n2-standard-64
disk_size_gb: 600
description: "64 vCPU, 256 GB RAM for large codebases"
EOF
Your workspace references an environment: 
# In workspace config
environment_ref: large

​
Placements

A placement (placement kind) specifies where VMs are provisioned:
FieldDescription
platformtrue = managed by Murmur (immutable to tenants), false = customer-managed
targetCloud-specific config: GCP (project, zones, subnet) or AWS (account, region, VPC, subnets)
substrateGCP or AWS
service_accountCloud identity for VMs
Platform placements are pre-configured by Murmur. They’re the default — you don’t need to create them. Your workspace references one by name. Customer placements are optional. You create them when you want agents running in your own cloud. See the Customer Placements guide.

​
Image families

Agent VMs use a two-layer image model:
  1. Platform base — Debian 12, Git, Node.js, Claude Code, gh CLI (maintained by Murmur)
  2. Your toolchain — Go, Python, Docker, CUDA, internal SDKs (customized via recipes)
Your workspace references an image: 
image_ref: go-python-toolchain
Create custom images with recipes and murmur bake. See the Custom VM Images guide.

​
Machine types

Machine types (machine-type kind) are pre-defined by the platform. List available ones: 
murmur get machine-type
Each machine type specifies vCPUs, memory, architecture (amd64/arm64), and optional GPU configuration.