executar-bugfix

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Chinese

Step 1: Context Analysis (Mandatory)

Read the bugs file at
```
./tasks/prd-[feature-slug]/bugs.md
```
and extract ALL documented bugs.
Read the PRD at
```
./tasks/prd-[feature-slug]/prd.md
```
to understand affected requirements.
Read the Tech Spec at
```
./tasks/prd-[feature-slug]/techspec.md
```
to understand relevant technical decisions.
Review project rules for compliance in fixes.
Do NOT skip this step — full context understanding is fundamental for quality fixes.

Step 2: Plan Fixes (Mandatory)

For each bug, generate a planning summary:
- Bug ID, Severity (High/Medium/Low), Affected Component.
- Root Cause analysis.
- Files to modify.
- Fix strategy description.
- Planned regression tests (unit, integration, E2E).
Use Context7 MCP to analyze documentation of involved languages, frameworks, and libraries.

Step 3: Implement Fixes (Mandatory)

Fix bugs in severity order: High first, then Medium, then Low.
For each bug follow this sequence: a. Locate and read the affected code. b. Reason about the flow causing the bug. c. Implement the root-cause fix — no superficial workarounds. d. Run
```
bun run typecheck
```
after each fix. e. Run existing tests to ensure no regressions.

Step 4: Create Regression Tests (Mandatory)

For each fixed bug, create tests that:
- Simulate the original bug scenario (test must fail if the fix is reverted).
- Validate the correct behavior with the fix applied.
- Cover related edge cases.
Choose test type based on bug nature:
- Unit test: Bug in isolated function/method logic.
- Integration test: Bug in module communication (e.g., controller + service).
- E2E test: Bug visible in the UI or full flow.

Step 5: Visual Validation with Playwright MCP (Mandatory for frontend bugs)

Step 6: Final Test Execution (Mandatory)

Step 7: Update bugs.md (Mandatory)

For each fixed bug, append to its entry:
- Status: Fixed.
- Applied fix: Brief description.
- Regression tests: List of created tests.

Step 8: Generate Final Report

步骤1：上下文分析（必做）

步骤2：制定修复计划（必做）

步骤3：执行修复（必做）

按严重程度顺序修复缺陷：优先修复高优先级，其次中优先级，最后低优先级。
每个缺陷的修复遵循以下流程： a. 定位并读取受影响的代码 b. 梳理导致缺陷的执行流程 c. 从根本原因层面实现修复，禁止使用表面的临时绕过方案 d. 每次修复后执行
```
bun run typecheck
```
e. 运行现有测试，确保没有引入回归问题

步骤4：编写回归测试（必做）

为每个已修复的缺陷编写测试用例，满足以下要求：
- 模拟原始缺陷场景（如果回滚修复，该测试必须失败）
- 验证修复应用后功能行为正确
- 覆盖相关边界场景
根据缺陷类型选择测试类型：
- 单元测试：孤立函数/方法逻辑存在的缺陷
- 集成测试：模块通信层面的缺陷（例如控制器 + 服务层）
- E2E测试：UI层面或全流程可见的缺陷

步骤5：使用Playwright MCP进行视觉验证（前端缺陷必做）

步骤6：执行最终测试（必做）

步骤7：更新bugs.md（必做）

步骤8：生成最终报告

If bugs.md does not exist, halt and report to the user.
If a bug requires significant architectural changes, document the justification before proceeding.
If new bugs are discovered during fixes, document them in bugs.md.
Begin implementation immediately after planning — do not wait for approval.