OpenClaw contributor
Contribute to OpenClaw the way the repo expects.
Start by reading the repo-root CONTRIBUTING.md in the target checkout. Treat it as the source of truth over generic PR habits.
Workflow
- 1. Confirm scope.
- Small bugfixes and focused docs fixes can go straight to a PR.
- New features, large refactors, or architecture changes should start with a GitHub Discussion or Discord conversation first.
- 2. Inspect the changed area before editing.
- Read nearby implementation and tests.
- Look for existing branch/work in
upstream/* before duplicating effort.
- 3. Generate a validation plan.
- Run
scripts/recommend_checks.py --repo <openclaw-repo>.
- Use its output to choose validation commands and maintainer routing hints.
- 4. Keep the patch tight.
- One logical change per PR.
- Add or update regression tests with the fix when possible.
- Avoid mixing runtime fixes, refactors, docs, and feature work in one branch.
- 5. Validate before opening the PR.
- Default expectation from OpenClaw is:
-
pnpm build
-
pnpm check
-
pnpm test
- For docs-only or subsystem-specific work, use the slimmer commands recommended by
scripts/recommend_checks.py.
- 6. Prepare the PR for maintainers.
- Explain what changed and why.
- Mark AI-assisted work in the PR title or description.
- State testing level clearly.
- Include screenshots for UI or visual changes.
- Optionally generate a draft with
scripts/generate_pr_body.py.
Non-negotiables
- - Follow
CONTRIBUTING.md, not generic habits. - Keep PRs focused.
- Prefer source-level fixes over patching built artifacts.
- Add tests for regressions when practical.
- Be transparent about AI assistance.
- For UI changes, preserve Control UI legacy decorator style unless the build tooling is intentionally being changed too.
Use bundled resources
- Read when you need the distilled OpenClaw-specific PR checklist, maintainer hints, or validation command matrix.
- Read when you need a maintainer-friendly OpenClaw PR structure with AI-assistance disclosure and validation sections.
- Run in an OpenClaw checkout to derive recommended validation commands and maintainer hints from the actual diff.
- Example:
-
python3 skills/openclaw-contributor/scripts/recommend_checks.py --repo /path/to/openclaw
-
python3 skills/openclaw-contributor/scripts/recommend_checks.py --repo /path/to/openclaw --base upstream/main --json
- Generate a PR-body draft using the diff-aware recommendations.
- Example:
- INLINECODE15
Related skills
If they are available locally, use them alongside this skill:
- -
github for GH CLI operations - INLINECODE17 before opening the PR
- INLINECODE18 before final submission or while addressing review comments
Output standard
When asked to contribute upstream, finish with:
- - branch name
- files changed
- validation run (or what remains and why)
- PR/discussion recommendation
- any maintainer/subsystem routing hints
技能名称: openclaw-contributor
详细描述:
OpenClaw 贡献者
按照仓库预期的方式为 OpenClaw 做出贡献。
首先阅读目标检出目录中仓库根目录下的 CONTRIBUTING.md 文件。将其视为超越通用 PR 习惯的权威依据。
工作流程
- 1. 确认范围。
- 小型错误修复和针对性的文档修复可以直接提交 PR。
- 新功能、大型重构或架构变更应首先通过 GitHub 讨论或 Discord 对话发起。
- 2. 在编辑前检查变更区域。
- 阅读附近的实现代码和测试。
- 在重复工作前，先查看 upstream/* 中是否存在现有分支/工作。
- 3. 生成验证计划。
- 运行 scripts/recommend_checks.py --repo
。
- 使用其输出来选择验证命令和维护者路由提示。
- 4. 保持补丁精简。
- 每个 PR 只包含一个逻辑变更。
- 尽可能在修复时添加或更新回归测试。
- 避免将运行时修复、重构、文档和功能工作混合在同一个分支中。
- 5. 在打开 PR 前进行验证。
- OpenClaw 的默认期望是：
- pnpm build
- pnpm check
- pnpm test
- 对于仅文档或子系统特定的工作，使用 scripts/recommend_checks.py 推荐的更精简命令。
- 6. 为维护者准备 PR。
- 解释变更内容和原因。
- 在 PR 标题或描述中标记 AI 辅助工作。
- 清晰说明测试级别。
- 对于 UI 或视觉变更，包含截图。
- 可选地使用 scripts/generateprbody.py 生成草稿。
不可妥协事项
- - 遵循 CONTRIBUTING.md，而非通用习惯。
- 保持 PR 聚焦。
- 优先使用源码级修复而非修补构建产物。
- 在实际可行时为回归问题添加测试。
- 对 AI 辅助保持透明。
- 对于 UI 变更，保留 Control UI 传统装饰器风格，除非构建工具也计划进行有意变更。
使用捆绑资源
- - references/contributing-checklist.md
- 当你需要精炼的 OpenClaw 特定 PR 检查清单、维护者提示或验证命令矩阵时阅读。
- - references/pr-template.md
- 当你需要包含 AI 辅助披露和验证部分的、对维护者友好的 OpenClaw PR 结构时阅读。
- - scripts/recommend_checks.py
- 在 OpenClaw 检出目录中运行，以从实际差异中推导推荐的验证命令和维护者提示。
- 示例：
- python3 skills/openclaw-contributor/scripts/recommend_checks.py --repo /path/to/openclaw
- python3 skills/openclaw-contributor/scripts/recommend_checks.py --repo /path/to/openclaw --base upstream/main --json
- - scripts/generateprbody.py
- 使用差异感知的推荐生成 PR 正文草稿。
- 示例：
- python3 skills/openclaw-contributor/scripts/generateprbody.py --repo /path/to/openclaw --title fix(web-search): honor OpenRouter-backed Perplexity runtime path --summary Honor OpenRouter-backed Perplexity config in runtime web_search path --why Current runtime ignores configured baseUrl and sends OpenRouter keys to Perplexity direct
相关技能
如果本地可用，请与此技能一同使用：
- - github 用于 GH CLI 操作
- Pull Request 在打开 PR 前使用
- code-review 在最终提交前或处理评审意见时使用
输出标准
当被要求向上游贡献时，以以下内容结束：
- - 分支名称
- 变更的文件
- 验证运行情况（或剩余工作及其原因）
- PR/讨论建议
- 任何维护者/子系统路由提示
