Maven Plugin & Lint

BLOGE ships two complementary build-time tools:

• bloge-maven-plugin for operator metadata export
• bloge-lint for static analysis of .bloge files

Together they make DSL assets and operator catalogs easier to validate and visualize in CI.

In 0.9.8-RC1, the same plugin also exposes Preview verify and verify-reactor goals backed by bloge-verification. They execute business Scenarios; they are distinct from static DSL lint.

Maven plugin: export operator metadata

The Maven plugin scans compiled operator classes, infers or reads their schema, and generates operator-metadata.json for schema-aware tooling and automation.

Add the plugin

<plugin>
  <groupId>com.leanowtech.bloge</groupId>
  <artifactId>bloge-maven-plugin</artifactId>
  <version>${bloge.version}</version>
  <executions>
    <execution>
      <goals>
        <goal>export-metadata</goal>
      </goals>
    </execution>
  </executions>
</plugin>

Run it manually

mvn bloge:export-metadata

What it exports

The generated JSON includes:

• operator name and class name
• input and output Java types
• inferred or explicit input schema
• inferred or explicit output schema
• generation metadata

This file becomes a machine-readable operator catalog for schema-aware tools.

Lint CLI and Maven integration

bloge-lint performs static analysis on .bloge files.

CLI usage

java -jar bloge-lint.jar check path/to/file.bloge
java -jar bloge-lint.jar check src/main/resources/bloge/

Common rules

Rule	Description
no-duplicate-node-id	Node ID declared more than once
no-duplicate-schema-name	Schema name declared more than once
no-unresolved-dependency	depends_on references a missing node
no-unresolved-branch-target	Branch target references a missing node
no-cycle	Graph contains a cycle
missing-timeout	Node has no timeout configured
missing-doc-comment	Node is missing documentation
excessive-fan-out	Node has too many outgoing edges

.blogerc.json

{
  "rules": {
    "missing-timeout": "warning",
    "missing-doc-comment": "off"
  }
}

Recommended CI flow

A common workflow looks like this:

1. compile operator classes
2. run bloge:export-metadata
3. run lint on committed .bloge assets
4. run bloge:verify when business-contract inputs and a test bootstrap are configured
5. publish metadata, verification artifacts, or repository outputs according to the selected gate

This keeps code-defined operators and externally authored DSL files aligned.

Preview verification goals

Compile test classes before loading the customer verification bootstrap:

mvn test-compile bloge:verify

For the supported direct-module reactor aggregate:

mvn test-compile bloge:verify-reactor

Keep the plugin version aligned through ${bloge.version}. A PASS from either goal is still separate from evidence trust and ClaimCapability; use the BLOGE Verify capability matrix before making it a release gate.

Using exported metadata

operator-metadata.json supports:

• operator catalog generation
• input/output contract inspection
• field completion and diagnostics
• automated documentation and compatibility checks

Practical guidance

• Generate metadata as part of your build, not as an afterthought.
• Treat lint failures as part of normal CI quality gates.
• Tune warning-level rules gradually as graph authoring discipline matures.
• Use metadata export to make operator contracts discoverable to non-runtime tooling.

Next steps

• Edit DSL safely with the VS Code Extension
• Explore schema contracts in Schema & I/O
• Configure business-contract execution in BLOGE Verify