AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified Apache-2.0 Self-run

Implement Code

skill-xinyiai0724-tools-implement-code · by xinyiai0724

在同一 CR workspace 内按 prd/sdd/tasks 执行代码编写,读取 CR worktree 文档并写入对应代码仓 CR worktree。

No reviews yet
0 installs
21 views
0.0% view→install

Install

$ agentstack add skill-xinyiai0724-tools-implement-code

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-xinyiai0724-tools-implement-code)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
1mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Implement Code? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Skill: implement-code

类型: 开发期 Skill(develop/ 组) 调用时机: code-implementation pipeline 的代码编写节点(approve-dev-start 后)


用途

在同一个 CR-YYYY-NNN workspace 内完成代码实现。代码与 CR 文档必须作为同一交付集合推进、checkpoint、评审和合并,不得单独推进文档或代码状态。启动前 CR status 必须已经由 approve-dev-start 推进到 developing


输入

| 参数 | 类型 | 必填 | 说明 | |------|------|------|------| | cr_id | string | yes | 目标 CR-ID | | runtime | string | no | 默认 openwork-opencode,不可用时 fallback 到第一个可用外部 CLI | | review_feedback | object | no | 来自 write-test-report 或 review-code 的 blockers、repair-instructions;存在时进入自修复模式 | | self_repair_attempt | number | no | 当前自动修复轮次,由 pipeline reviewLoop 注入 |


CR Workspace Contract

  1. 先读取 dir-graph.yaml#repositories,解析 active repo,不得硬编码某个固定平台目录或 code
  2. 文档根固定来自 knowledge-base CR worktree:.xinyiai-worktrees/knowledge-base/requirement/{cr_id}
  3. 独立 git 代码仓写入 .xinyiai-worktrees/{repo.id}/requirement/{cr_id}
  4. 非独立代码目录写入 knowledge-base CR worktree 内对应路径。
  5. 主工作区 change-requests/、平台运行时代码目录、主工作区 code/ 不参与开发期读写。

必读文档

从 CR worktree 读取:

  • {crWorkspaceDocsRoot}/change-requests/{cr_id}/prd.md
  • {crWorkspaceDocsRoot}/change-requests/{cr_id}/sdd.md
  • {crWorkspaceDocsRoot}/change-requests/{cr_id}/tasks/TASK-*.md

缺少任一文件或任一 active repo worktree 不存在时,立即停止并返回结构化错误。


执行步骤

Step 1 - Resolve Repo Map

调用 CR workspace resolver,根据 dir-graph.yaml#repositories 生成 repo map:

  • role=knowledge-base 使用 bucket knowledge-base
  • 独立 git 代码仓使用 bucket {repo.id}
  • 非独立代码目录使用 knowledge-base worktree 内的 {repo.path}

Step 2 - Validate Workspace

校验:

  • CR 当前 status 为 developing
  • 所有 active repo 的 CR worktree 存在
  • 所有参与仓当前分支为 requirement/{cr_id}
  • PRD、SDD、TASK 文件来自 CR worktree

Step 3 - Implement Tasks

  • 优先使用目标运行时已安装的 external subagent-driven-development 执行 TASK。
  • 不支持子 agent 时使用目标运行时已安装的 external executing-plans
  • 每个 TASK 必须遵循目标运行时已安装的 external test-driven-development:先写失败测试,再实现,再验证转绿。
  • 实现只写 repo map 指定的 codeRoot。
  • 若存在 review_feedback,进入自修复模式:读取 blockers、repair-instructions、repair-target 与上一轮 test-report.md / review-annotations/code.yml,只修复被指出的问题,避免无关重构,并输出 fixed-blockers。

Step 4 - Verify

按 TASK 验收条件运行 lint/test/build 或对应验证命令。任何失败都必须记录到节点输出并停止推进;在自修复模式下至少重新运行与 blockers 相关的验证命令,并说明未运行全量验证的原因。

Step 5 - Output

节点输出必须包含:

  • 实际读取的 PRD/SDD/TASK 路径
  • 实际写入的 repo/worktree/codeRoot
  • runtime 与 fallback 情况
  • run/session id
  • 验证命令与结果
  • 未完成项或失败原因
  • 自修复模式下的 reviewfeedback 摘要、selfrepair_attempt 与 fixed-blockers

禁止事项

  • 禁止读取或写入主工作区 change-requests/
  • 禁止在平台运行时代码目录或主工作区 code/ 直接编码。
  • 禁止只提交或只推送文档、只提交或只推送代码。
  • 禁止在缺少任一参与 repo worktree 时继续执行。

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.