Doc Architecture

SkillDocs & knowledge

Integrates requirements discovery, domain modeling, and consistency auditing into the /team-* main chain, producing auditable documentation with artifacts/adr/memory as the sole persistence targets. Use when the Tech Lead, Architect, or QA needs to complete or evolve architecture documentation.

Use Doc Architecture in Claude, ChatGPT or Ahel Desktop

Free. Sign in, add Doc Architecture and connect your AI. About a minute.

Also: Claude Code · Cursor · Codex

Then ask your AI: use the Doc Architecture skill

Details

Instructions available. Your AI can read the instructions. Execution depends on the setup they require.

Add Ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.

Doc ArchitectureStart free

What this skill tells your AI

The instructions your AI receives, as published by hashgraph-online/awesome-codex-plugins in plugins/Colin4k1024/tsp/skills/doc-architecture/SKILL.md and read by Ahel’s review.

用途

  • 将文档能力拆入现有 team-skills 主链,不新增并行文档体系。
  • 统一 Discovery、Service Decomposition、NFR、Audit 的输入输出结构。
  • 保证文档可追溯落盘到 docs/artifacts/、docs/adr/ 与 docs/memory/。

完整工作流

Step 1: Discovery

  1. 使用 discovery-questionnaire.md 收集项目背景、技术栈、架构风格与约束。
  2. 若是既有项目(brownfield),先补做现状扫描:盘点现有模块、入口、外部依赖、关键数据流和历史包袱;必要时借助 /update-codemaps 生成 token-lean 结构快照。
  3. 形成 Project Profile Card。
  4. 按 tech-stack-profiles.md 锁定文档术语风格。

输出回落:prd.md、delivery-plan.md

Step 2: Modeling

  1. 使用 service-decomposition-guide.md 识别领域与服务边界。
  2. 产出 Service Catalog 与 Communication Matrix。
  3. 收集 NFR(可用性、性能、安全、可观测性)。

输出回落:arch-design.md、api-contract.md

Step 3: Generation Mapping

  1. 按 artifact-mapping.md 将文档输出映射到 artifacts。
  2. 参考 references/templates/ 选择章节结构。
  3. 若属于 brownfield 任务,把现状扫描结果回落为 brownfield context snapshot,写入 delivery-plan.md / arch-design.md,不要新建平行事实源。
  4. 对缺失信息使用 <!-- TODO: ... --> 标记,不虚构业务事实。

输出回落:delivery-plan.md、arch-design.md、api-contract.md

Step 4: Execute Backfill

  1. 在执行阶段记录实现偏差、接口漂移、事件变更。
  2. 关键取舍进入 ADR,轻量决策进入 decisions。

输出回落:execute-log.md、docs/adr/*.md、docs/memory/decisions.md

Step 5: Consistency Audit

  1. 使用 audit-checklist.md 做一致性审计。
  2. 审计服务名、API 覆盖、事件覆盖、鉴权一致性、索引完整性。

输出回落:test-plan.md

Step 6: Release & Evolution

  1. 将发布、监控、回滚与观察项落入发布产物。
  2. 按 doc-evolution-guide.md 做增量演进。

输出回落:release-plan.md、docs/memory/sessions/*.md

最小产物清单

  1. docs/artifacts/{date}-{slug}/prd.md
  2. docs/artifacts/{date}-{slug}/delivery-plan.md
  3. docs/artifacts/{date}-{slug}/arch-design.md
  4. docs/artifacts/{date}-{slug}/api-contract.md(按需)
  5. docs/artifacts/{date}-{slug}/execute-log.md
  6. docs/artifacts/{date}-{slug}/test-plan.md
  7. docs/artifacts/{date}-{slug}/release-plan.md
  8. docs/adr/ADR-{NNN}-{slug}.md(按需)
  9. docs/memory/project-context.md、docs/memory/decisions.md、docs/memory/lessons-learned.md、docs/memory/sessions/*.md

何时优先调用

  • tech-lead 需要把架构文档能力并入主链而不新增命令时。
  • architect 需要在 arch-design.md 与 api-contract.md 形成可执行契约时。
  • qa-engineer 需要把文档一致性纳入放行证据时。

参考资料

Signals

GitHub stars
1k
Forks
316
Last commit
Oct 2026
Advanced
Item type
skill
Key
doc-architecture
Source
github.com/hashgraph-online/awesome-codex-plugins