Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

Latest commit


Failed to load latest commit information.

Blast Radius

CircleCI PyPI version

Blast Radius is a tool for reasoning about Terraform dependency graphs with interactive visualizations.

Use Blast Radius to:

  • Learn about Terraform or one of its providers through real examples
  • Document your infrastructure
  • Reason about relationships between resources and evaluate changes to them
  • Interact with the diagram below (and many others) in the docs



Note: For macOS you can brew install graphviz


The fastest way to get up and running with Blast Radius is to install it with pip to your pre-existing environment:

pip install blastradius

Once installed just point Blast Radius at any initialized Terraform directory:

blast-radius --serve /path/to/terraform/directory

And you will shortly be rewarded with a browser link


To launch Blast Radius for a local directory by manually running:

docker run --rm -it -p 5000:5000 \
  -v $(pwd):/data:ro \
  --security-opt apparmor:unconfined \
  --cap-add=SYS_ADMIN \

A slightly more customized variant of this is also available as an example docker-compose.yml usecase for Workspaces.

Docker configurations

Terraform module links are saved as absolute paths in relative to the project root (note .terraform/modules/<uuid>). Given these paths will vary betwen Docker and the host, we mount the volume as read-only, assuring we don't ever interfere with your real environment.

However, in order for Blast Radius to actually work with Terraform, it needs to be initialized. To accomplish this, the container creates an overlayfs that exists within the container, overlaying your own, so that it can operate independently. To do this, certain runtime privileges are required -- specifically --cap-add=SYS_ADMIN.

For more information on how this works and what it means for your host, check out the runtime privileges documentation.

Docker & Subdirectories

If you organized your Terraform project using stacks and modules, Blast Radius must be called from the project root and reference them as subdirectories -- don't forget to prefix --serve!

For example, let's create a Terraform project with the following:

$ tree -d
`-- project/
    |-- modules/
    |   |-- foo
    |   |-- bar
    |   `-- dead
    `-- stacks/
        `-- beef/
             `-- .terraform

It consists of 3 modules foo, bar and dead, followed by one beef stack. To apply Blast Radius to the beef stack, you would want to run the container with the following:

$ cd project
$ docker run --rm -it -p 5000:5000 \
    -v $(pwd):/data:ro \
    --security-opt apparmor:unconfined \
    --cap-add=SYS_ADMIN \
    28mm/blast-radius --serve stacks/beef

Embedded Figures

You may wish to embed figures produced with Blast Radius in other documents. You will need the following:

  1. An svg file and json document representing the graph and its layout.
  2. javascript and css found in .../blastradius/server/static
  3. A uniquely identified DOM element, where the <svg> should appear.

You can read more details in the documentation

Implementation Details

Blast Radius uses the [Graphviz][] package to layout graph diagrams, PyHCL to parse Terraform configuration, and d3.js to implement interactive features and animations.

Further Reading

The development of Blast Radius is documented in a series of blog posts:

  • part 1: motivations, d3 force-directed layouts vs. vanilla graphviz.
  • part 2: d3-enhanced graphviz layouts, meaningful coloration, animations.
  • part 3: limiting horizontal sprawl, supporting modules.
  • part 4: search, pan/zoom, prune-to-selection, docker.

A catalog of example Terraform configurations, and their dependency graphs can be found here.

These examples are drawn primarily from the examples/ directory distributed with various Terraform providers, and aren't necessarily ideal. Additional examples, particularly demonstrations of best-practices, or of multi-cloud configurations strongly desired.