architecture-maplisted

# Architecture Map ## Activation Contract Use this skill when the user wants visual architecture documentation for a project: system context, containers, and key runtime flows. Do not use it for code-level design review, user story mapping (`usm`), or AI-harness analysis (`absorb`). ## Hard Rules - C4-lite: System Context (level 1) and Container (level 2) only. Component-level diagrams only on explicit request. - Every node and edge comes from code evidence: manifests, build files, configs, routes, entrypoints, deployment descriptors. Never from README claims. Unverifiable elements are marked `hypothesis`. - Derive containers, dependencies, and flows from imports, configs, deployment descriptors, or a code-graph index (for example, CodeGraph MCP/CLI) when available, before file-by-file search; a graph-resolved edge still cites the underlying `file:line`. - Budgets: at most ~30 nodes per diagram and ~120 lines per doc. Split flows into more diagrams before exceeding a budget. - Mermaid only (`flowchart`, `sequenceDiagram`), GitHub-renderable. No external image tooling. - Visual first: diagrams carry the weight; prose is one short paragraph per diagram. ## Doc Folder Rule Write under `<docfolder>/architecture/`, where `<docfolder>` is the project's existing `docs/`, else its existing `doc/`, else a newly created `doc/`. ## Doc Set Three files; concrete skeletons live in `assets/architecture-doc-set.md`: - `index.md`: ~10 lines — what each doc answers, generation date,