grepwrapper

TotalClaw 作者 totalclaw

通过 grepwrapper CLI 使用 grep.app 在公共 GitHub 存储库中搜索精确的代码匹配。

安装 / 下载方式

TotalClaw CLI推荐
totalclaw install totalclaw:totalclaw~riprsa-grepwrapper
cURL直接下载,无需登录
curl -fsSL https://skills.taituai.com/api/skills/totalclaw%3Atotalclaw~riprsa-grepwrapper/file -o riprsa-grepwrapper.md
# grepwrapper

Use this skill when the user asks to:
- find exact code matches on GitHub
- locate repo/file/path occurrences for a snippet
- run grep.app code search from CLI

## Install

```bash
npm i -g git+https://github.com/riprsa/grepwrapper.git
```

## Standard usage

```bash
grepwrapper search --q "PendingBalanceAt(ctx context.Context, account common.Address) (*big.Int, error)"
```

Options:
- `--q <query>` required
- `--case` case-sensitive
- `--words` whole-word (mutually exclusive with `--regexp`)
- `--regexp` regex mode (mutually exclusive with `--words`)
- `--page <n>` page number

Examples:

```bash
grepwrapper search --q "QMD" --case --regexp
grepwrapper search --q "QMD" --case --words --page 2
```

## Expected output

CLI returns summary lines:
- `time=<ms> total=<n> returned=<n>`
- one line per hit: `- <repo>:<path> (matches=<count>)`

## Agent behavior

- Return top matches with direct GitHub links when possible.
- If many matches exist, summarize and ask whether user wants next page / narrower filters.
- Prefer exact query first, then widen with `--regexp` only when needed.

---

## 中文说明

# grepwrapper

在用户要求执行以下操作时使用此技能:
- 在 GitHub 上查找精确的代码匹配
- 定位某段代码片段出现的存储库/文件/路径
- 从命令行运行 grep.app 代码搜索

## 安装

```bash
npm i -g git+https://github.com/riprsa/grepwrapper.git
```

## 标准用法

```bash
grepwrapper search --q "PendingBalanceAt(ctx context.Context, account common.Address) (*big.Int, error)"
```

选项:
- `--q <query>` 必填
- `--case` 区分大小写
- `--words` 全词匹配(与 `--regexp` 互斥)
- `--regexp` 正则模式(与 `--words` 互斥)
- `--page <n>` 页码

示例:

```bash
grepwrapper search --q "QMD" --case --regexp
grepwrapper search --q "QMD" --case --words --page 2
```

## 预期输出

CLI 返回摘要行:
- `time=<ms> total=<n> returned=<n>`
- 每个命中一行:`- <repo>:<path> (matches=<count>)`

## 代理行为

- 尽可能返回靠前的匹配结果,并附带直达的 GitHub 链接。
- 如果存在大量匹配,进行汇总,并询问用户是否需要下一页/更精确的过滤条件。
- 优先使用精确查询,仅在需要时再用 `--regexp` 扩大范围。