Skip to main content

Diff visualizer for Databricks bundle plans — resources and job DAGs

Project description

Dagshund

Colored diff summaries and interactive DAG visualizations for Databricks Asset Bundles. Requires a plan file from the direct deployment engine (databricks bundle plan -o json). See what changed, what's new, and what's being deleted, in your terminal or in the browser.

Dagshund resource graph

Install

Requires Python 3.12+.

# Run without installing (ephemeral)
uvx dagshund plan.json

# Persistent install
uv tool install dagshund

# Traditional pip
pip install dagshund

Usage

By default, dagshund prints a colored text diff to the terminal:

dagshund plan.json

Terminal output

Export an interactive HTML visualization with -o:

dagshund plan.json -o report.html

Open it in the browser automatically with -b:

dagshund plan.json -o report.html -b

Reads from stdin, so you can pipe directly from the Databricks CLI:

databricks bundle plan -o json | dagshund
databricks bundle plan -o json | dagshund -o report.html

Dagshund works anywhere, it just needs a plan JSON file. You don't need to run it from inside your bundle directory.

Filter the output to specific change types:

dagshund plan.json -c              # changed resources only (hides unchanged)
dagshund plan.json -a              # added only
dagshund plan.json -m              # modified only
dagshund plan.json -a -r           # added and removed

The filter flags (-a, -m, -r) compose freely. -c is shorthand for -a -m -r.

Use -f to filter by resource type, name, or diff status with the same search DSL as the browser UI:

dagshund plan.json -f 'type:jobs'               # only jobs
dagshund plan.json -f 'status:added'             # only new resources
dagshund plan.json -f '"etl_pipeline"'            # exact name match
dagshund plan.json -f 'type:jobs pipeline'       # jobs matching "pipeline"
dagshund plan.json -c -f 'type:volumes'          # changed volumes only

All tokens in a filter expression AND together. -f composes with -c/-a/-m/-r — both must match.

Use -e for CI-friendly exit codes (see CI Exit Codes).

Interactive Visualization

The HTML report shows your resources as an interactive graph with diff highlighting. Resources are organized into visual groups:

  • Workspace, jobs, alerts, experiments, pipelines, and other bundle resources
  • Unity Catalog, catalogs, schemas, volumes, and registered models in their hierarchy
  • Lakebase, database instances and synced tables (when present, flat resources move into an "Other Resources" group)

Click any node to open a detail panel with per-field structural diffs, old values in red, new values in green, unchanged fields for context.

Structural diff detail panel

Jobs with task dependencies get their own DAG view. Switch between the Resources and Jobs tabs to navigate between them.

Job task DAG

Phantom Nodes

If your schema references a parent catalog that isn't in your bundle, dagshund infers it and adds it to the graph as a phantom node (shown with a dashed border). Hierarchy phantoms like these always display because they hold the tree together. Inferred leaf nodes (like a warehouse referenced by an alert) can be toggled on or off with the Inferred leaf nodes button in the toolbar.

Phantom nodes

Lateral Dependencies

Many resources reference each other across hierarchies, an alert might target a SQL warehouse, or a serving endpoint might bind to a registered model. These relationships are hidden by default to keep the graph clean. Toggle Lateral dependencies in the toolbar to see how your resources connect across group boundaries.

Lateral dependencies

Search

The search bar uses the same filter DSL as the CLI's -f flag. Non-matching nodes dim so matches stand out.

  • Type a name to filter: warehouse, analytics
  • Wrap in quotes for exact match: "analytics" finds only that node, not analytics_pipeline
  • Prefix with type: to filter by badge: type:wheel highlights all wheel tasks
  • Prefix with status: to filter by diff status: status:added, status:modified, status:removed
  • Press Escape to clear

The diff filter buttons (Added, Modified, Removed) compose with search, when both are active, only nodes matching both criteria stay highlighted. When exactly one node matches, the viewport auto-centers on it.

CI Exit Codes

Use --detailed-exitcode (or -e) for machine-readable exit codes in CI pipelines:

dagshund plan.json -e
Exit Code Meaning
0 Plan parsed, no changes detected
1 Error (bad input, missing file, etc.)
2 Plan parsed, changes detected
# Check for drift
dagshund plan.json -e
if [ $? -eq 2 ]; then echo "Drift detected"; fi

# Generate report AND get exit code
dagshund plan.json -o report.html -e

Without -e, dagshund always exits 0 on success.

Agent Skill

Dagshund ships as a Claude Code plugin with an agent skill that teaches AI coding agents how to use it. Once installed, your agent can answer questions like "what's changing in my deploy?" by running dagshund automatically.

Install as a Claude Code plugin:

/plugin marketplace add https://github.com/chinchyisbored/dagshund.git
/plugin install dagshund@chinchy-dagshund

Or use the CLI to install the skill into any agent harness:

dagshund --install-skill .claude/skills     # Claude Code
dagshund --install-skill .cursor/skills     # Cursor
dagshund --install-skill .agents/skills     # Codex / Gemini CLI

Contributing

Dagshund is a solo project and I'm not accepting pull requests at this time. If you run into a bug or have a feature request, please open an issue, I'm happy to hear what you need.

Development

Requires Bun 1.3+, uv, and just.

just install        # Install dependencies
just dev            # Dev server with hot reload (http://localhost:3000)
just build          # JS template + Python wheel
just check          # Lint + typecheck + all tests

For Claude Code LSP support (go-to-definition, type hover, find-references), install the language server binaries:

bun install -g pyright typescript-language-server typescript

The plugins are already configured in .claude/settings.json.

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

dagshund-0.5.2.tar.gz (1.6 MB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

dagshund-0.5.2-py3-none-any.whl (763.4 kB view details)

Uploaded Python 3

File details

Details for the file dagshund-0.5.2.tar.gz.

File metadata

  • Download URL: dagshund-0.5.2.tar.gz
  • Upload date:
  • Size: 1.6 MB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: uv/0.10.4 {"installer":{"name":"uv","version":"0.10.4","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Debian GNU/Linux","version":"13","id":"trixie","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}

File hashes

Hashes for dagshund-0.5.2.tar.gz
Algorithm Hash digest
SHA256 d76a2a7c3d59e55a013758b34d99e3d521bd401b3aaeda07bcdc3fd5c5f2ba17
MD5 95690788ef6c86ec8d0cde446905c7a0
BLAKE2b-256 b4487c2823b460a8417c6ecbb09848c21a5278aeb96675c3df09061d1632c5b1

See more details on using hashes here.

File details

Details for the file dagshund-0.5.2-py3-none-any.whl.

File metadata

  • Download URL: dagshund-0.5.2-py3-none-any.whl
  • Upload date:
  • Size: 763.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: uv/0.10.4 {"installer":{"name":"uv","version":"0.10.4","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Debian GNU/Linux","version":"13","id":"trixie","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}

File hashes

Hashes for dagshund-0.5.2-py3-none-any.whl
Algorithm Hash digest
SHA256 b934adcf913a4376765d41df63b51f6043e6c19d78c6c2ed77e027f3cd5ba2d0
MD5 401229659a3c38aa305ff282774e43ee
BLAKE2b-256 6afaced8f0a28cd0edbe7c83b616aee56ad6a1393f2df0e7ec712e81ef543622

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page