What problem does it solve? Turning a PRD, epic, or idea into sound technical direction is hard: teams either jump straight into task lists or over-architect. This Skill runs a structured working session that explores multiple approaches, weighs trade-offs, and records the high-level architecture decisions before any implementation planning begins. ## Core Features & Use Cases - Interactive exploration loop: Investigates the intent, surfaces 2-3 approaches with trade-offs, recommends a direction with reasoning, and waits for the user's call before going deeper. - Greenfield and brownfield modes: Explores the solution space for new builds, or reads the existing codebase first to decide how new work plugs into an existing system. - Architecture decision doc output: Produces a high-level decision document covering approaches considered, stack and libraries, data model, boundaries, missing pieces, spikes, and open questions — as a linked tracker page (Confluence/Jira via Atlassian MCP), a section in the PRD, or a standalone file. - Use Case: Given a Jira epic for a new notifications feature, run the skill to compare event-driven vs polling approaches, pick a stack, flag a spike for delivery reliability, and write the decision doc as a Confluence page linked to the epic. ## Quick Start Run the plan-architecture skill with the path to your PRD or epic, for example asking to explore architecture options for docs/prd-notifications.md and produce a decision document.
