跳至正文
OpenAgentSkill
浏览已开放 · 目录记录日期 2026-11-21 · 自动更新和账户服务暂时关闭。
开发与测试

feature

Macro workflow for building or changing a vertical slice of the CLI. Use as the entry point when adding a new end-to-end feature (domain → use-case → adapter → command → tests) or when a change touches multiple layers at once. Do NOT use for single-layer changes — use the layer skill directly (`domain-model`, `use-case`, `adapter`, `command`, or `test`).

ai-driven-dev · 已有 AI 审核记录

安装与使用

运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。

npx skills add ai-driven-dev/framework --skill feature

技能说明


name: feature description: > Macro workflow for building or changing a vertical slice of the CLI. Use as the entry point when adding a new end-to-end feature (domain → use-case → adapter → command → tests) or when a change touches multiple layers at once. Do NOT use for single-layer changes — use the layer skill directly (domain-model, use-case, adapter, command, or test).

Feature

Coordinates the five layer skills in vertical-slice order. Each step delegates entirely to the relevant layer skill. Skip any step when the change does not touch that layer.

Available actions

#ActionRoleInput
01domain-modelDefine types, value objects, and invariantsconcept description
02use-caseImplement business orchestrationdomain types from 01
03adapterAdd I/O boundary (only if a new port is needed)use-case port needs
04commandExpose the feature in the CLIuse-case from 02
05testWrite pyramid coverageall layers from 01-04

Default flow

01 → 02 → 03 → 04 → 05

Skip rule: if a step's layer is not affected by the change, skip it explicitly and document why (e.g. "03 skipped — no new port needed, reusing existing PluginFetcher").

Conditional layers

These layers are triggered only when the change explicitly touches their domain. Evaluate each before starting the main flow and apply them in parallel with whichever main steps they overlap.

LayerTrigger conditionSkill
toolAdding or modifying an AI tool definition in domain/tools/ai/tool
formatAdding or modifying a pure string-transform function in domain/formats/format
capabilityAdding or modifying a capability class in domain/capabilities/capability
  • If the feature adds a new AI tool: run tool before step 01 (the tool definition underpins the domain model).
  • If the feature adds a pure format transform: run format at the same level as step 01.
  • If the feature adds a capability class: run capability before tool (the Has* interface must exist before the tool composes it).
  • All three may be skipped when the change does not touch their respective domains — document the skip explicitly.

Transversal rules

  • Each action delegates fully to its layer skill. Do not inline layer-specific rules here.
  • Never skip 05 — every feature change requires tests at the appropriate pyramid tier.
  • Skipping 01 is allowed only when no new domain type is introduced and no existing invariant changes.
  • Skipping 03 is the most common skip — only add an adapter when a genuinely new I/O boundary is required.
  • Skipping 04 is allowed for internal refactors that don't expose a new CLI surface.

External data

  • .claude/skills/domain-model/SKILL.md — layer skill for step 01
  • .claude/skills/use-case/SKILL.md — layer skill for step 02
  • .claude/skills/adapter/SKILL.md — layer skill for step 03
  • .claude/skills/command/SKILL.md — layer skill for step 04
  • .claude/skills/test/SKILL.md — layer skill for step 05
  • .claude/skills/tool/SKILL.md — conditional layer skill for AI tool definitions
  • .claude/skills/format/SKILL.md — conditional layer skill for pure string transforms
  • .claude/skills/capability/SKILL.md — conditional layer skill for capability classes