piv-implement

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Execute: Implement from Plan

执行:从计划到实施

Plan to Execute

待执行计划

Read plan file:
$ARGUMENTS
读取计划文件:
$ARGUMENTS

Before you start — work on a feature branch

开始前——在功能分支上工作

A ticket gets built on its own branch, so it can become one PR. Ideally you're already on that branch — cut it before planning — so the plan commit you made is on it and rides into the PR; a plan committed on the base branch won't be in this branch's PR. If you're still on base, this step creates the branch now. Detect the base branch (don't hardcode
main
):
git symbolic-ref refs/remotes/origin/HEAD 2>/dev/null | sed 's@^refs/remotes/origin/@@'
(fallback
main
).
  • On the base branch, clean → create one:
    git checkout -b feature/<plan-slug>
    .
  • Already on a feature branch or in a worktree → use it.
  • On the base branch with uncommitted changes → STOP: commit or stash first.
(One branch per ticket is also what makes parallel worktrees clean later.)
每个工单都在独立分支上开发,最终可合并为一个PR。理想情况下你已经处于该分支——在规划前就创建好分支——这样你提交的计划会包含在分支中并进入PR;如果计划提交在基础分支上,那么它不会出现在此分支的PR中。 如果你仍在基础分支上,此步骤会立即创建分支。自动检测基础分支(不要硬编码
main
):
git symbolic-ref refs/remotes/origin/HEAD 2>/dev/null | sed 's@^refs/remotes/origin/@@'
(默认 fallback 为
main
)。
  • 当前在基础分支且工作区干净 → 创建分支:
    git checkout -b feature/<plan-slug>
  • 已在功能分支或工作树中 → 直接使用当前分支。
  • 当前在基础分支且有未提交的更改 → 停止:先提交或暂存更改。
(每个工单对应一个分支也能让后续并行工作树保持整洁。)

Execution Instructions

执行说明

1. Read and Understand

1. 阅读与理解

  • Read the ENTIRE plan carefully
  • Understand all tasks and their dependencies
  • Note the validation commands to run
  • Review the testing strategy
  • 仔细阅读完整计划
  • 理解所有任务及其依赖关系
  • 记录需要运行的验证命令
  • 审核测试策略

2. Execute Tasks in Order

2. 按顺序执行任务

For EACH task in "Step by Step Tasks":
针对“分步任务”中的每一项任务

a. Navigate to the task

a. 定位任务

  • Identify the file and action required
  • Read existing related files if modifying
  • 明确需要操作的文件及动作
  • 如果是修改操作,先阅读相关现有文件

b. Implement the task

b. 实现任务

  • Follow the detailed specifications exactly
  • Maintain consistency with existing code patterns
  • Include proper type hints and documentation
  • Add structured logging where appropriate
  • 严格遵循详细规格要求
  • 与现有代码模式保持一致
  • 添加适当的类型提示和文档
  • 在合适的位置添加结构化日志

c. Verify as you go

c. 逐步验证

  • After each file change, check syntax
  • Ensure imports are correct
  • Verify types are properly defined
  • Run the task's own
    VALIDATE
    command before starting the next task.
    Every task in the plan carries one. A task is not done until its check passes — if it fails, fix it now rather than carrying the failure forward. The full suite still runs at step 4; this is the per-task gate that keeps step 4 from becoming a pile-up.
  • 每次文件修改后,检查语法正确性
  • 确保导入语句正确
  • 验证类型定义无误
  • 在开始下一项任务前,运行当前任务对应的
    VALIDATE
    命令。
    计划中的每个任务都配有验证命令。只有验证通过,任务才算完成——如果验证失败,立即修复,不要带着问题继续推进。步骤4仍会运行完整验证套件,但这一步的单任务验证能避免步骤4出现大量问题堆积。

3. Implement Testing Strategy

3. 落实测试策略

After completing implementation tasks:
  • Create all test files specified in the plan
  • Implement all test cases mentioned
  • Follow the testing approach outlined
  • Ensure tests cover edge cases
完成所有实施任务后:
  • 创建计划中指定的所有测试文件
  • 实现计划中提及的所有测试用例
  • 遵循计划中概述的测试方法
  • 确保测试覆盖边缘场景

4. Run Validation Commands

4. 运行验证命令

Execute ALL validation commands from the plan in order:
bash
undefined
按顺序执行计划中的所有验证命令:
bash
undefined

Run each command exactly as specified in plan

严格按照计划中的指定命令执行


If any command fails:
- Fix the issue
- Re-run the command
- Continue only when it passes

如果任何命令失败:
- 修复问题
- 重新运行命令
- 只有验证通过后才能继续

5. Final Verification

5. 最终验证

Before completing:
  • ✅ All tasks from plan completed
  • ✅ All tests created and passing
  • ✅ All validation commands pass
  • ✅ Code follows project conventions
  • ✅ Documentation added/updated as needed
完成前确认:
  • ✅ 计划中的所有任务已完成
  • ✅ 所有测试已创建并通过
  • ✅ 所有验证命令执行通过
  • ✅ 代码符合项目规范
  • ✅ 已按需添加/更新文档

Output — write an implementation report

输出——撰写实施报告

Write a short report to
.claude/reports/<plan-slug>-report.md
(and print the summary). This is what the PR body and the
piv-review-pr
gate read — especially the deviations (a documented deviation is an intentional decision the reviewer should not flag):
markdown
undefined
将简短报告写入
.claude/reports/<plan-slug>-report.md
(并打印摘要)。这份报告是PR正文和
piv-review-pr
环节的依据——尤其要注意偏差说明(有文档记录的偏差是评审者无需标记的 intentional 决策):
markdown
undefined

Implementation Report — <feature>

实施报告 —— <功能名称>

Plan: <path> Branch: <feature/...> Status: COMPLETE | PARTIAL
计划:<路径> 分支:<feature/...> 状态:已完成 | 部分完成

Summary

摘要

{What was built, 2-4 sentences.}
{已完成的内容,2-4句话。}

Tasks completed

已完成任务

  • [task] →
    path/to/file
    (CREATE/UPDATE)
  • [任务内容] →
    path/to/file
    (创建/更新)

Tests added

添加的测试

{Test files + cases + results.}
{测试文件 + 用例 + 结果。}

Validation results

验证结果

{Type-check / lint / tests / build — pass/fail with counts.}
{类型检查 / 代码检查 / 测试 / 构建 —— 结果(通过/失败)及数量。}

Deviations from the plan

与计划的偏差

{What changed vs the plan and WHY — or "none". This is the reviewer's signal of intent.}
{与计划不同的修改及原因 —— 或“无偏差”。这是给评审者的意图信号。}

Issues encountered

遇到的问题

{Anything notable, or "none".}
undefined
{任何值得注意的问题 —— 或“无”。}
undefined

Ready for the next step

准备进入下一步

  • Confirm all changes are complete and validations pass.
  • Next:
    piv-commit
    the work, then
    piv-create-pr
    to open the PR (the report fills the PR body), then
    piv-review-pr
    .
  • 确认所有修改已完成且验证通过。
  • 下一步:
    piv-commit
    提交工作,然后
    piv-create-pr
    创建PR(报告将填充PR正文),接着执行
    piv-review-pr

Notes

注意事项

  • If you encounter issues not addressed in the plan, document them
  • If you need to deviate from the plan, explain why
  • If tests fail, fix implementation until they pass
  • Don't skip validation steps
  • 如果遇到计划中未提及的问题,请记录下来
  • 如果需要偏离计划,请说明原因
  • 如果测试失败,修复实现直到测试通过
  • 不要跳过验证步骤