code-execution

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Code Execution

代码执行

Overview

概述

Execute coding subtasks with self-review and quality validation. Runs in isolated coder-agent context with pre-loaded standards.

Announce at start: "I'm using the code-execution skill to implement [subtask title]."

执行带有自我审查和质量验证的编码子任务。在预加载了标准的独立coder-agent环境中运行。

开始时声明： "我正在使用code-execution技能来实现[子任务标题]。"

The Process

流程

Step 1: Read Subtask JSON

步骤1：读取子任务JSON

Load the subtask file specified in $ARGUMENTS:

bash

Read: .tmp/tasks/{feature}/subtask_{seq}.json

Extract:

```
title
```
— What to implement
```
acceptance_criteria
```
— What defines success
```
deliverables
```
— Files/endpoints to create
```
context_files
```
— Standards to apply
```
reference_files
```
— Existing code to study

加载$ARGUMENTS中指定的子任务文件：

bash

Read: .tmp/tasks/{feature}/subtask_{seq}.json

提取以下内容：

```
title
```
— 要实现的内容
```
acceptance_criteria
```
— 成功的定义
```
deliverables
```
— 要创建的文件/端点
```
context_files
```
— 要应用的标准
```
reference_files
```
— 要参考的现有代码

Step 2: Load Context Files

步骤2：加载上下文文件

Read each file in

context_files

bash

Read: .opencode/context/core/standards/code-quality.md
Read: .opencode/context/core/standards/security-patterns.md

Understand:

Project coding standards
Naming conventions
Security patterns
Code quality requirements

读取

context_files

中的每个文件：

bash

Read: .opencode/context/core/standards/code-quality.md
Read: .opencode/context/core/standards/security-patterns.md

理解以下内容：

项目编码标准
命名规范
安全模式
代码质量要求

Step 3: Load Reference Files

步骤3：加载参考文件

Read each file in

reference_files

bash

Read: src/middleware/auth.middleware.ts

Study:

Existing patterns
Code structure
Conventions in use

读取

reference_files

中的每个文件：

bash

Read: src/middleware/auth.middleware.ts

研究以下内容：

现有模式
代码结构
当前使用的规范

Step 4: Update Status to In Progress

步骤4：将状态更新为进行中

Edit subtask JSON:

json

"status": "in_progress",
"agent_id": "coder-agent",
"started_at": "2026-02-16T00:00:00Z"

编辑子任务JSON：

json

"status": "in_progress",
"agent_id": "coder-agent",
"started_at": "2026-02-16T00:00:00Z"

Step 5: Implement Deliverables

步骤5：实现交付物

For each deliverable:

Create or modify the specified file
Follow acceptance criteria EXACTLY
Apply standards from context_files
Use patterns from reference_files
Write clean, modular, functional code

针对每个交付物：

创建或修改指定文件
严格遵循验收标准
应用context_files中的标准
使用reference_files中的模式
编写清晰、模块化、功能完善的代码

Step 6: Run Self-Review (MANDATORY)

步骤6：执行自我审查（必填）

Type & Import Validation:

✅ Function signatures match usage
✅ All imports/exports exist
✅ No missing type annotations
✅ No circular dependencies

Anti-Pattern Scan:

bash

grep "console.log" deliverables  # NO debug statements
grep "TODO\|FIXME" deliverables  # NO unfinished work
grep "api[_-]key\|secret" deliverables  # NO hardcoded secrets

Acceptance Criteria Check:

Re-read
```
acceptance_criteria
```
array
Confirm EACH criterion met
Fix unmet criteria before proceeding

类型与导入验证：

✅ 函数签名与使用匹配
✅ 所有导入/导出均存在
✅ 无缺失的类型注解
✅ 无循环依赖

反模式扫描：

bash

grep "console.log" deliverables  # 禁止调试语句
grep "TODO\|FIXME" deliverables  # 禁止未完成工作
grep "api[_-]key\|secret" deliverables  # 禁止硬编码密钥

验收标准检查：

重新阅读
```
acceptance_criteria
```
数组
确认每个标准都已满足
继续之前修复未满足的标准

Step 7: Mark Complete

步骤7：标记完成

Update subtask status:

bash

bash .opencode/skills/task-management/router.sh complete {feature} {seq} "{summary}"

Verify:

bash

bash .opencode/skills/task-management/router.sh status {feature}

更新子任务状态：

bash

bash .opencode/skills/task-management/router.sh complete {feature} {seq} "{summary}"

验证：

bash

bash .opencode/skills/task-management/router.sh status {feature}

Step 8: Return Report

步骤8：返回报告

✅ Subtask {feature}-{seq} COMPLETED

Self-Review: ✅ Types | ✅ Imports | ✅ No debug code | ✅ Criteria met

Deliverables:
- src/auth/jwt.service.ts
- src/auth/jwt.service.test.ts

Summary: JWT service with RS256 signing and 15min token expiry

✅ 子任务 {feature}-{seq} 已完成

自我审查：✅ 类型检查 | ✅ 导入检查 | ✅ 无调试代码 | ✅ 符合验收标准

交付物：
- src/auth/jwt.service.ts
- src/auth/jwt.service.test.ts

摘要：支持RS256签名和15分钟令牌过期时间的JWT服务

Error Handling

错误处理

Missing subtask JSON:

Main agent must create subtask before invoking this skill

Context files not found:

Main agent must run
```
/context-discovery
```
first

Acceptance criteria unmet:

DO NOT mark complete—fix issues first

缺失子任务JSON：

主Agent必须先创建子任务，再调用此技能

未找到上下文文件：

主Agent必须先运行
```
/context-discovery
```

验收标准未满足：

禁止标记完成——先修复问题

Red Flags

危险信号

If you think any of these, STOP and re-read this skill:

"I know the pattern, I don't need to read the context files"
"I'll do the self-review quickly at the end"
"The acceptance criteria are obvious, I don't need to check them"
"I'll mark it done and fix the edge cases later"

如果你有以下想法，请停止并重新阅读本技能：

"我知道这个模式，不需要读上下文文件"
"我会在最后快速完成自我审查"
"验收标准很明显，我不需要检查"
"我会先标记完成，之后再修复边缘情况"

Common Rationalizations

常见自我合理化借口

Excuse	Reality
"I've seen this pattern before, context files will just confirm what I know"	Projects diverge from common patterns. One wrong assumption = rework. Read the files.
"Self-review is just checking my own work"	Self-review catches type errors, missing imports, and debug code before the main agent sees it.
"The criteria are implied by the task title"	Implied criteria are unverifiable. If it's not written, it's not a gate.
"I'll handle edge cases in a follow-up"	Edge cases left for follow-up become bugs in production. Handle them now.

借口	实际情况
"我之前见过这个模式，上下文文件只会确认我知道的内容"	项目会偏离通用模式。一个错误的假设就会导致返工。请务必阅读文件。
"自我审查只是检查自己的工作"	自我审查能在主Agent看到之前捕获类型错误、缺失导入和调试代码。
"验收标准从任务标题就能看出来"	隐含的标准无法验证。没有书面记录的内容，不能作为交付门槛。
"我会在后续处理边缘情况"	留到后续处理的边缘情况会变成生产环境中的Bug。现在就处理它们。

Remember

注意事项

Context FIRST, code SECOND—always read context_files before implementing
One subtask at a time—complete fully before moving on
Self-review is MANDATORY—quality gate before marking done
Functional, declarative, modular—clean code patterns only
Return results to main agent—report completion for orchestration

先看上下文，再写代码——实施前务必阅读context_files
一次处理一个子任务——完全完成后再进行下一个
自我审查是必填项——标记完成前的质量关卡
仅使用功能完善、声明式、模块化的清晰代码模式
向主Agent返回结果——报告完成情况以进行编排

code-execution

Original

Translation

Code Execution

代码执行

Overview

概述

The Process

流程

Step 1: Read Subtask JSON

步骤1：读取子任务JSON

Step 2: Load Context Files

步骤2：加载上下文文件

Step 3: Load Reference Files

步骤3：加载参考文件

Step 4: Update Status to In Progress

步骤4：将状态更新为进行中

Step 5: Implement Deliverables

步骤5：实现交付物

Step 6: Run Self-Review (MANDATORY)

步骤6：执行自我审查（必填）

Step 7: Mark Complete

步骤7：标记完成

Step 8: Return Report

步骤8：返回报告

Error Handling

错误处理

Red Flags

危险信号

Common Rationalizations

常见自我合理化借口

Remember

注意事项

Related

相关技能