orvanta dmn

Manage DMN decision tables in your workspace. Decisions are addressed by path (e.g. f/demo/loan_approval), the same way flows and scripts are.

Note

DMN decisions are a Business-tier, not-yet-shipped capability — see Decision models (DMN). These commands talk to the same API the product UI would use, so they 404/402 against an instance where the feature isn’t built or entitled.

Usage

orvanta dmn <subcommand> [options]

Local file layout

A DMN decision is kept as two files on disk: the raw, byte-exact {path}.dmn XML, plus a {path}.dmn_decision.json/.yaml sidecar carrying summary, description, and labels — metadata that doesn’t exist in DMN 1.3 XML at all. The XML is never embedded inside the sidecar, so it round-trips byte-for-byte through git.

Subcommands

new (alias bootstrap)

Create a new DMN decision locally — writes both the .dmn file (a minimal empty decision table) and its sidecar. Doesn’t touch the workspace.

orvanta dmn new f/demo/loan_approval --summary "Loan approval"

list

List DMN decisions in the workspace.

orvanta dmn list [--show-archived] [--json]

get

Show a decision’s metadata, or just its raw XML.

orvanta dmn get f/demo/loan_approval
orvanta dmn get f/demo/loan_approval --xml

push

Push a local .dmn file, overriding the remote version. This deploys — it does not go through the draft table.

orvanta dmn push ./loan_approval.dmn f/demo/loan_approval --message "Add REVIEW band"

Pushing the sidecar file directly is rejected — point push at the raw .dmn file; the sidecar next to it (if present) is picked up automatically.

eval

Evaluate a decision — either one already saved by path, or a local file with --file.

orvanta dmn eval f/demo/loan_approval --data '{"age":30,"country":"NL","amount":10000}'
orvanta dmn eval --file ./loan_approval.dmn --data '{"age":30,"country":"NL","amount":10000}'

• -d/--data — inputs as a JSON string, @<filename>, or @-/- for stdin.
• --decision-ref <id> — selects a specific <decision id> when the document defines more than one.
• --json — print the full EvaluateResponse (hit policy, matched rules, outputs, and any unmapped inputs) as JSON.
• -s/--silent — print only the outputs object, for scripting.
• --fail-on-no-match — exit 1 when no rule matched, instead of the default exit 0. “No rule matched” is a normal result (Value::Null), not an error, unless you opt into this.

The human-readable output (no --json/--silent) prints the hit policy, matched rule numbers, a table of output values, and — if any input had no FEEL representation and was silently excluded from evaluation — which one(s), the same warning the decision-table editor’s test panel shows.

history / show-version

Version history for a decision, or one specific version’s XML.

orvanta dmn history f/demo/loan_approval
orvanta dmn show-version f/demo/loan_approval 12

archive / delete

orvanta dmn archive f/demo/loan_approval
orvanta dmn delete f/demo/loan_approval

What’s not here

There is no orvanta dmn service (or any decision-service/DRD subcommand) — Decision Services aren’t a shipped surface yet. There is also no deploy subcommand distinct from push/eval --file: pushing a .dmn file always deploys it.

Related

• Decision models (DMN): hit policies, the FEEL cell grammar, and engine limits.
• The decision-table editor: the same operations from the product UI.
• Sync: deploy a .dmn/sidecar pair alongside the rest of a workspace.
• Build your first decision: a step-by-step tutorial using this CLI and the editor together.