菜单
Architecture visualization patterns using Mermaid and ASCII diagrams. Use for service flows, sequence diagrams, C4 models, decision matrices, or expressing architectural ideas visually.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Sprint planning patterns including effort estimation, cross-project validation, ticket refinement workflows, two-validator parallel intent-only refinement, Foundation→Gates ticket pairs, workstream split thresholds, and hub/per-project reconciliation. Use for sprint capacity planning, ticket validation, dependency discovery, tier-gated dispatch, strategic pivots, or coordinating multi-agent ticket processing.
Multimodal AI patterns for extracting structured information from video using Gemini, GPT-4o, and Claude vision. Frame sampling strategies, structured output with Pydantic, temporal reasoning, and cost optimization. Use for video analysis, report generation from video, and visual inspection.
Apple Foundation Models framework for on-device AI. LanguageModelSession, @Generable macro, structured output, token management, and availability checking.
Tactical DDD patterns for clean Python architecture including Domain Model, Repository, Service Layer, Unit of Work, and Aggregates. Use for backend architecture decisions, implementing data access layers, separating business logic from infrastructure, or reviewing code separation.
Architecture Decision Records (ADRs) for documenting significant technical decisions. Use for capturing decision context, rationale, consequences, and maintaining decision history.
Event-driven architecture patterns for Python including Domain Events, Commands, Message Bus, CQRS, and Bootstrap. Use for implementing event sourcing, decoupled workflows, reactive systems, microservices integration, or making decisions about event-driven design.
| name | arch-diagrams |
| description | Architecture visualization patterns using Mermaid and ASCII diagrams. Use for service flows, sequence diagrams, C4 models, decision matrices, or expressing architectural ideas visually. |
| allowed-tools | Read, Grep, Glob |
| kb-sources | ["wiki/software-engineering/arch-diagrams"] |
| updated | "2026-04-18T00:00:00.000Z" |
Patterns for expressing architectural ideas visually.
| Need to Show | Use | Tool |
|---|---|---|
| Service dependencies | Flowchart | Mermaid |
| Request/response timing | Sequence | Mermaid |
| System boundaries | C4 Context | Mermaid |
| Internal architecture | C4 Container | Mermaid |
| Quick inline sketch | ASCII Box | Text |
| Option comparison | Decision Matrix | Table |
Mermaid Flowchart: flowchart TD → A[Node] --> B[Node]
Mermaid Sequence: sequenceDiagram → A->>B: Message
C4 Model: C4Context or C4Container → Person(), System(), Container()
ASCII Box: ┌───┐ │ │ └───┘ with ───▶ arrows
Use ASCII art for spatial layouts (UI wireframes, page structure) and Mermaid for sequential flows (user journeys, data flow). Each tool applied to its strength — ASCII shows position, Mermaid shows order.
See reference.md for complete syntax, examples, and advanced patterns.
Tables are a distinct visualization tool from diagrams, and picking the right shape changes how fast cold readers parse a system.
| Table shape | What it reveals | Use for |
|---|---|---|
| Taxonomic (flat category list) | Completeness of a set | Listing alternatives, feature matrices |
| Sequenced (temporal ordering) | "What comes when" | Disambiguating workflow commands, phased rollouts, lifecycle stages |
| Artifacts (producer → consumer) | Hidden dependency chains and silent failure modes | Multi-command workflows, data pipelines, agent handoffs |
Sequenced tables disambiguate better than taxonomic ones when commands or concepts feel interchangeable. Reframing three ambiguous options as "session → weekly review → housekeeping" clicks faster than a flat category list.
Artifacts tables (3 columns: artifact / producer / consumers) make it explicit when skipping one step silently degrades downstream steps. For workflow docs where commands hand state off to each other, this is often the single most useful table in the document — it surfaces failure modes that prose buries. Pair with a sequence diagram (Mermaid) when the temporal order also matters.
