Forge: JIRA Ticket → Dynamic Skill → Implementation

Inputs

Raw arguments: $ARGUMENTS

Infer from the arguments:

• TICKET_URL: the JIRA ticket URL
• ADDITIONAL_CONTEXT: any extra context provided beyond the URL (optional)

Orchestrates end-to-end JIRA ticket processing. Phases 1–5 gather context and write output files into a dynamically created skill at ~/.claude/skills/{ticket_id}/. Phase 6 invokes that skill — context files are lazy-loaded on demand.

1. Initialization — create skill directory at ~/.claude/skills/{ticket_id}/
2. JIRA Analysis (via Skill("jira-ticket-viewer"))
3. Discovery & Scanning (all 3.x run concurrently):
   • 3.1 Domain Discovery (via Skill("domain-discover"))
   • 3.2 Resource Scanning (links, Figma designs)
   • 3.3 Page Inspection (conditional, via Skill("page-inspector"))
4. Prompt Optimization (via Skill("meta-prompter"))
5. Implementation Planning + generate SKILL.md for the dynamic skill
6. Execute — invoke Skill("{ticket_id}-impl")

Also check the ADDITIONAL_CONTEXT as dev server info hint if provided.

Skill Dependencies

The following skills are invoked during orchestration:

• Skill("jira-ticket-viewer") — Fetch JIRA ticket details via jira CLI
• Skill("confluence-page-viewer") — Read Confluence pages via confluence CLI
• Skill("figma-reader") — Read Figma designs (when Figma links present in ticket)
• Skill("domain-discover") — Domain knowledge discovery
• Skill("page-inspector") — Capture current page layout/styles as baseline (conditional: frontend/UI-affecting changes)
• Skill("meta-prompter") — Prompt evaluation and optimization
• Skill("{ticket_id}-impl") — Dynamically generated skill for execution (created in Phase 5)

Execution

Follow references/phases.md for step-by-step phase instructions.