cmd-speckit-plan

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

/speckit.plan

Agent skill wrapper for the Claude command

/speckit.plan

When the original command text references

{{INPUT}}

$1

, or named arguments, map them from the user's current request.

Claude命令

/speckit.plan

的Agent技能封装。

当原始命令文本引用

{{INPUT}}

、

$1

或命名参数时，从用户当前请求中映射对应内容。

Command Instructions

命令说明

Produce a complete technical implementation plan from a feature specification.

根据功能规格文档生成完整的技术实施方案。

Steps

步骤

Locate the spec:
- If input names a feature or file path, use it.
- Otherwise check
```
.specify/specs/*/spec.md
```
  ; if multiple, ask which to plan.
- Fall back to
```
spec.md
```
  in the project root.
- Set
```
FEATURE_DIR
```
  to the spec's parent directory.

Load context: Read

spec.md

and

.specify/memory/constitution.md

. Detect tech stack from

package.json

mix.exs

Cargo.toml

pyproject.toml

go.mod

, etc.

Phase 0 — Research: For each
```
[NEEDS CLARIFICATION]
```
marker and technical unknown:
- Search the codebase for existing patterns and abstractions.
- Identify required external dependencies.
- Write findings to
```
FEATURE_DIR/research.md
```
  :
  markdown
```
# Research: <Feature Name>
## Unknowns Investigated
## Existing Patterns Found
## External Dependencies
```
- Resolve all open questions before proceeding.

Phase 1 — Design artifacts:

Write

FEATURE_DIR/plan.md

markdown

# Technical Plan: <Feature Name>
## Architecture Overview
## Technology Choices (with rationale)
## Component Design
## Implementation Phases
## Risk Assessment
## Open Technical Questions

Write

FEATURE_DIR/data-model.md

(only if the feature involves persistent data):

markdown

# Data Model: <Feature Name>
## Entities (fields, types, constraints, relationships)
## Migrations (in dependency order)
## Indexes (with justification)

Create

FEATURE_DIR/contracts/<interface-name>.md

for each external interface:

markdown

# Contract: <Interface Name>
## Signature / Endpoint
## Request / Input Schema
## Response / Output Schema
## Error Conditions
## Constraints

Constitution check: Verify the plan does not violate any MUST or SHOULD NOT principle. If violations found, flag
```
[CONSTITUTION CONFLICT]
```
and pause for user review before continuing.
Report: List all files created with paths and one-line summaries.

Next step: Run

/speckit.analyze

to validate consistency, or

/speckit.tasks

to generate the task breakdown.

定位规格文档:
- 如果输入指定了功能名称或文件路径，则使用该路径。
- 否则检查
```
.specify/specs/*/spec.md
```
  ；如果存在多个，询问用户要为哪个生成方案。
- 若以上都不存在，则使用项目根目录下的
```
spec.md
```
  。
- 将
```
FEATURE_DIR
```
  设置为规格文档的父目录。

加载上下文：读取

spec.md

和

.specify/memory/constitution.md

。从

package.json

、

mix.exs

、

Cargo.toml

、

pyproject.toml

、

go.mod

等文件中检测技术栈。

阶段0 — 调研：针对每个
```
[NEEDS CLARIFICATION]
```
标记和技术疑点：
- 在代码库中搜索现有模式和抽象实现。
- 确定所需的外部依赖。
- 将调研结果写入
```
FEATURE_DIR/research.md
```
  :
  markdown
```
# Research: <Feature Name>
## Unknowns Investigated
## Existing Patterns Found
## External Dependencies
```
- 在进入下一阶段前解决所有未明确的问题。

阶段1 — 设计产出物:

编写

FEATURE_DIR/plan.md

markdown

# Technical Plan: <Feature Name>
## Architecture Overview
## Technology Choices (with rationale)
## Component Design
## Implementation Phases
## Risk Assessment
## Open Technical Questions

编写

FEATURE_DIR/data-model.md

（仅当功能涉及持久化数据时）:

markdown

# Data Model: <Feature Name>
## Entities (fields, types, constraints, relationships)
## Migrations (in dependency order)
## Indexes (with justification)

为每个外部接口创建

FEATURE_DIR/contracts/<interface-name>.md

markdown

# Contract: <Interface Name>
## Signature / Endpoint
## Request / Input Schema
## Response / Output Schema
## Error Conditions
## Constraints

合规性检查：验证方案是否违反任何MUST或SHOULD NOT原则。若发现违规，标记
```
[CONSTITUTION CONFLICT]
```
并暂停流程，等待用户审核后再继续。
生成报告：列出所有创建的文件，包含路径和单行摘要。

下一步：运行

/speckit.analyze

验证一致性，或运行

/speckit.tasks

生成任务分解。