brin-check

分类: 工具与效率 | 上传者: superagent-ai | 下载: 0 | 版本: v1.0（最新）
使用 brin API 扫描包、存储库、MCP 服务器、域、网页和代理技能是否存在安全威胁。在安装依赖项、访问 URL 或集成外部资源之前使用此技能。
更新日志: Source: GitHub https://github.com/superagent-ai/brin
目录结构

当前层级: tree/main/plugins/cursor/skills/brin-check/
📄 SKILL.md 5.8 KB
SKILL.md

---
name: brin-check
description: Scan packages, repositories, MCP servers, domains, web pages, and agent skills for security threats using the brin API. Use this skill before installing dependencies, visiting URLs, or integrating external resources.
---

# brin Security Check

Scan any external resource for security threats using the brin API before installing, visiting, or using it.

## Usage

When the user asks to check or scan a resource, when you need to verify the safety of an external dependency, or when you encounter unfamiliar URLs or domains:

1. Identify the resource type and name
2. Call the appropriate brin API endpoint
3. Report the verdict, score, and any detected threats

## API Reference

The brin API requires no authentication. Make a GET request to:

```
https://api.brin.sh/<origin>/<name>
```

### Supported Origins

| Origin | Name | Example |
|--------|------|---------|
| `npm` | package name | `https://api.brin.sh/npm/express` |
| `pypi` | package name | `https://api.brin.sh/pypi/requests` |
| `crate` | crate name | `https://api.brin.sh/crate/serde` |
| `domain` | domain | `https://api.brin.sh/domain/example.com` |
| `page` | domain/path | `https://api.brin.sh/page/example.com/path` |
| `repo` | owner/repo | `https://api.brin.sh/repo/facebook/react` |
| `mcp` | owner/repo | `https://api.brin.sh/mcp/modelcontextprotocol/servers` |
| `skill` | owner/repo | `https://api.brin.sh/skill/owner/skill-name` |

### Response Format

The API returns JSON with these fields:

| Field | Type | Description |
|-------|------|-------------|
| `origin` | string | Source type (e.g. `npm`, `pypi`, `domain`) |
| `name` | string | Resource name |
| `score` | integer | 0–100 safety score (higher is safer) |
| `confidence` | string | `low`, `medium`, or `high` |
| `verdict` | string | `safe`, `caution`, `suspicious`, or `malicious` |
| `tolerance` | string | Scan tolerance level |
| `scanned_at` | string | ISO 8601 timestamp |
| `url` | string | API endpoint URL for this resource |

Response headers also provide quick access: `x-brin-verdict`, `x-brin-score`, and `x-brin-confidence`.

## Step-by-Step

### 1. Determine the origin and identifier

Parse the user's request to determine what type of resource they want to check. Examples:
- "Check if lodash is safe" → `npm/lodash`
- "Scan the requests Python package" → `pypi/requests`
- "Is github.com/owner/repo safe?" → `repo/owner/repo`
- "Check this MCP server: owner/server" → `mcp/owner/server`
- "Is example.com safe?" → `domain/example.com`
- "Check this URL: https://example.com/path" → `page/example.com/path`
- "Scan the domains from my web search results" → `domain/<each-domain>`

### 2. Query the API

```bash
curl -s https://api.brin.sh/npm/lodash
curl -s https://api.brin.sh/domain/example.com
curl -s https://api.brin.sh/page/example.com/some/path
```

### 3. Interpret the results

Report the following to the user:

- **Verdict**: The overall safety assessment
- **Score**: The numerical score (0–100)
- **Confidence**: How confident brin is in the result

### 4. Recommend action

| Verdict | Recommendation |
|---------|---------------|
| `safe` | Safe to use. Proceed with installation. |
| `caution` | Generally safe but review the flagged concerns before proceeding. |
| `suspicious` | Significant risk detected. Recommend finding an alternative. |
| `malicious` | Do NOT install. This resource contains known threats. |

## Example Output

When reporting results, format them clearly:

```
brin scan: express (npm)
━━━━━━━━━━━━━━━━━━━━━━━━
Score:      84/100
Verdict:    safe ✓
Confidence: medium
```

For a caution result:

```
brin scan: requests (pypi)
━━━━━━━━━━━━━━━━━━━━━━━━
Score:      65/100
Verdict:    caution ⚠
Confidence: medium
```

For a malicious resource:

```
brin scan: malicious-pkg (npm)
━━━━━━━━━━━━━━━━━━━━━━━━
Score:      12/100
Verdict:    malicious ✗
Confidence: high

⚠ Do NOT install this package.
```

## Batch Scanning

When multiple packages are being installed, scan each one individually and provide a summary table:

```
brin scan summary
━━━━━━━━━━━━━━━━━━━━━━━━
Package     Score  Verdict
express     84     safe ✓
lodash      88     safe ✓
requests    65     caution ⚠
```

## Web Search and URL Scanning

When web search results return URLs or the user provides links, scan the domains before visiting:

### Scanning domains from web search

```bash
curl -s https://api.brin.sh/domain/example.com
```

### Scanning a specific page

```bash
curl -s https://api.brin.sh/page/example.com/docs/guide
```

### Example output for a domain scan

```
brin scan: example.com (domain)
━━━━━━━━━━━━━━━━━━━━━━━━
Score:      90/100
Verdict:    safe ✓
Confidence: high
```

For a dangerous domain:

```
brin scan: malicious-site.xyz (domain)
━━━━━━━━━━━━━━━━━━━━━━━━
Score:      8/100
Verdict:    malicious ✗
Confidence: high

⚠ Do NOT visit this domain.
```

### When to scan URLs

- After a web search returns results with unfamiliar domains
- Before following links from external documentation or READMEs
- When a user asks you to visit or read content from a URL
- Before scraping or extracting data from any web page
- When evaluating third-party API documentation links

### Batch URL scanning

For multiple URLs from search results, scan each domain and provide a summary:

```
brin domain scan summary
━━━━━━━━━━━━━━━━━━━━━━━━
Domain              Score  Verdict
docs.example.com    95     safe ✓
blog.trusted.io     88     safe ✓
free-tools.xyz      15     suspicious ⚠
```
登录后下载/点赞/收藏 ❤ 35 | ★ 0
brin-check

目录结构

SKILL.md

举报内容

提示
