Implement AI-powered workflow analysis using Google Gemini to transform raw Jira issue hierarchies into structured, QE-actionable insights. The feature enables test teams to create test plans directly from customer workflows without manual research.

  Key Capabilities

  | Capability          | Description                                                    |
  |--------------------|---------------------------------------------------------------|
  | Multimodal Analysis | Process text, images, and PDFs in single Gemini API call       |
  | Component Detection | Identify RHOAI components (KServe, Pipelines, Notebooks, etc.) |
  | Adoption Tracking   | Classify workflows as requested/implemented/adopted            |
  | QE Output Format    | Structured repro steps, prerequisites, expected artifacts      |
  | Per-Issue Mode      | Generate separate insights for each linked issue in epics      |

  QE-Aligned Output Fields

  workflow_name, workflow_description, area, personas, rhoai_version,
  entry_point, prerequisites, repro_steps, expected_artifacts,
  visualization, implementation_status, adoption_confidence, source_issues

  Commands

• Single issue analysis

  uv run python main.py analyze --issue RHAIRFE-297

•  Epic mode (per-linked-issue insights)

  uv run python main.py analyze --epic CIPOE-148887

•  Force regeneration

  uv run python main.py analyze --issue RHAIRFE-297 --force

Integration Points

  - Input: Sanitized extraction JSON from data/issues/{KEY}/extraction.json
  - Attachments: Images/PDFs from data/issues/{KEY}/attachments/
  - API: Google Gemini (gemini-3-flash default)
  - Output: insight.json + insight.md to Central Issue Pool

Acceptance Criteria

  - Insights include repro steps usable by QE without additional research
  - All visual attachments (PNG, JPG, PDF) included in analysis
  - RHOAI components correctly identified from workflow context
  - Output validates against Pydantic schema
  - Smart caching skips unchanged issues (0 API calls)