一键导入
system-design
System architecture discipline — decomposition, boundaries, communication, scale, failure design. Load for architectural work; decisions land as ADRs.
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。
菜单
System architecture discipline — decomposition, boundaries, communication, scale, failure design. Load for architectural work; decisions land as ADRs.
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
Context engineering — memory navigation, textual graph traversal, progressive disclosure, recovery. Load when assembling context for a task or when lost.
Implementation-side security discipline — headers, dependencies, secrets, authz. Load for hardening work (sec-ops domain).
Commit conventions, branching, anchors in messages, and the git safety rules. Load for any git operation beyond status/diff.
Anti-generic-AI design discipline — bans, tells, and the slop test. Load for ANY new UI surface or visual redesign. Inspired by Impeccable (Apache 2.0).
Motion & animation discipline — purpose, performance, reduced-motion. Load when work involves animation, transitions or scroll effects.
Microcopy discipline — buttons, errors, empty/loading/success states, alt text. Voice and tone come from the project's memory, not from this skill.
| name | system-design |
| description | System architecture discipline — decomposition, boundaries, communication, scale, failure design. Load for architectural work; decisions land as ADRs. |
The area-level companion to the architect role — sr-back proposes with this discipline, architect ratifies via ADR. Infra counterpart: platform-architecture.
Sync (request/response) for queries the caller must wait on · async events (see async-jobs) for facts other contexts react to · eventual consistency is a product decision — surface it to the human ("the count may lag by seconds — acceptable?"), never silently assume.
Contracts are versioned and additive-first (see api-design §versioning); consumers tolerate unknown fields.
Measure first (see observability — p95/p99, not averages) → stateless services scale out; state is the real bottleneck → push state to stores built for it → cache deliberately (see caching) → estimate with numbers before architecting: back-of-envelope QPS/storage/growth in the ADR. Optimizing for imagined scale fails review; so does ignoring stated scale requirements from the PRD.
Everything fails: every remote call has a timeout · retries with budgets (idempotency prerequisite — see async-jobs) · circuit breakers on flaky dependencies · graceful degradation designed per feature (what does the user see when X is down?) · bulkheads: one dependency's failure must not exhaust shared resources (pools, threads).
Architectural proposals carry: the forcing function · the cost accepted · back-of-envelope numbers · the failure story · rollback path. Goes to deliberation when alternatives are genuine; ADR always; human decides (never fast-path — this is the architectural tier by definition).