What problem does it solve?
Architecture diagrams are often drawn from memory or assumptions, producing visuals that misrepresent the real system. This Skill builds diagrams strictly from repository evidence—code, runtime wiring, IaC, and contracts—so every element and relationship is traceable and trustworthy.
Core Features & Use Cases
- Evidence-Backed Modeling: Traces components, containers, stores, queues, and dependencies from actual repository artifacts, marking uncertain relationships as UNKNOWN instead of guessing.
- Multiple View Types: Produces system-context, container, component, sequence, data-flow, and deployment views, selecting only the minimum views that answer a named audience question.
- Verifiable Output: Validates Mermaid or ASCII syntax, keeps diagram source reviewable in version control, and reports a READY, INCONCLUSIVE, or BLOCKED verdict with a completion checklist.
- Use Case: When onboarding engineers to a microservices system, generate a current-state container diagram with labeled relationships and evidence notes so the team can verify each claim against the codebase.
Quick Start
Ask the agent to create an evidence-backed architecture diagram of the current system for a specific audience and question, saving it under docs/architecture/diagrams.