content-review-pipeline

Content Review Pipeline

Content Review Pipeline -- AI generates draft, human reviews/edits, then publish.

Input: topic, audience | Timeout: 300s

Output: content, wordCount, model, published, url

Pipeline

crp_ai_draft
    │
crp_human_review [WAIT]
    │
crp_publish

Workers

CrpAiDraftWorker (crp_ai_draft): Worker for crp_ai_draft task -- generates an AI content draft.

• topic: the content topic
• audience: the target audience
• content: generated draft content string incorporating topic and audience
• wordCount: 42 (deterministic.word count)

Reads audience, topic. Outputs content, wordCount, model.

CrpPublishWorker (crp_publish): Worker for crp_publish task -- publishes content if approved.

• approved: boolean indicating whether the content was approved
• content: the content to publish
• published: true if approved and published, false otherwise
• url: the published URL (only present when published=true)

boolean approved = Boolean.TRUE.equals(approvedObj);

Reads approved. Outputs published, url.

Workflow Output

• content: ${crp_ai_draft_ref.output.content}
• wordCount: ${crp_ai_draft_ref.output.wordCount}
• model: ${crp_ai_draft_ref.output.model}
• published: ${crp_publish_ref.output.published}
• url: ${crp_publish_ref.output.url}

Data Flow

crp_ai_draft: topic = ${workflow.input.topic}, audience = ${workflow.input.audience} crp_human_review [WAIT]: content = ${crp_ai_draft_ref.output.content}, wordCount = ${crp_ai_draft_ref.output.wordCount}, model = ${crp_ai_draft_ref.output.model} crp_publish: approved = ${crp_human_review_ref.output.approved}, content = ${crp_human_review_ref.output.editedContent}

Tests

16 tests cover valid inputs, boundary values, null handling, and error paths.

mvn test

---

Run this example: see RUNNING.md for setup, build, and CLI instructions.