This server is an MCP security audit tool for pull requests. It analyzes code diffs to detect security risks and generates review comments.
Tools:
analyze_pr_diff: Submits a unified diff text and returns a structured JSON report of security findings.
analyze_github_pr: Fetches and analyzes a GitHub PR diff by URL (supports optional token for private repos).
generate_review_comment: Generates a Markdown security audit comment from a diff, ready for PR review.
Detection capabilities:
Secret leaks: AWS keys, GitHub tokens, connection strings, private key fragments.
High-risk patterns: eval, command execution, SQL string concatenation.
New dependencies: Scans package.json, requirements.txt, pyproject.toml, pom.xml.
Output: Structured JSON with a severity summary (critical, high, medium, low) and detailed findings (category, severity, file, line, evidence, fix suggestions).
Integration: Works with MCP clients like Codex, Claude Code, Cursor.
Limitations: Uses heuristic regex (not full SAST); does not automatically post comments; dependency detection only identifies new packages without vulnerability lookup.
Allows analysis of GitHub pull requests by fetching diffs and detecting security risks such as secrets, dangerous patterns, and new dependencies.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@pr-guard-mcp analyze this diff for security issues"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
pr-guard-mcp 是一个面向 Pull Request 的安全审计 MCP Server。
它可以让 Codex、Claude Code、Cursor 等支持 MCP 的 AI Agent 对 git diff 或 PR diff 做安全检查，帮助你在代码合并前发现明显风险。
检测新增代码中的密钥泄露，例如 AWS key、GitHub token、数据库连接串、私钥片段
识别高风险代码模式，例如 eval、命令执行、SQL 字符串拼接
识别依赖文件里的新增依赖，例如 package.json、requirements.txt、pyproject.toml、pom.xml
输出结构化 JSON，方便 AI Agent 生成 PR review 评论
Related MCP server: CodePeel MCP Server
npm install
npm run buildnpm run devnpm run demo示例会分析 examples/insecure.diff，输出类似下面的安全摘要：
{
"summary": {
"totalFindings": 3,
"critical": 1,
"high": 1,
"medium": 1
},
"findings": [
{
"category": "dangerous-pattern",
"severity": "critical",
"title": "Potential command execution added"
}
]
}analyze_pr_diff分析一段 unified diff 文本，并返回安全风险列表。
输入：
{
"diff": "这里放 git diff 或 PR diff 文本"
}输出字段：
summary：按严重程度统计的风险数量
findings：具体风险列表，包括类型、严重程度、文件、行号、证据和修复建议
analyze_github_pr输入 GitHub PR URL，自动拉取 PR diff 并进行安全分析。
输入：
{
"prUrl": "https://github.com/owner/repo/pull/123",
"token": "可选，私有仓库或高频请求时使用"
}说明：
公共仓库可以不传 token
私有仓库需要传 GitHub token
token 只用于请求 GitHub API，不会出现在分析结果里
generate_review_comment输入 diff 文本，生成可以直接粘贴到 PR 里的 Markdown 安全审计评论。
输入：
{
"diff": "这里放 git diff 或 PR diff 文本"
}输出示例：
## PR 安全审计结果
发现 1 个风险：critical 1，high 0，medium 0，low 0。
| 严重程度 | 类型 | 位置 | 问题 | 建议 |
|---|---|---|---|---|
| critical | dangerous-pattern | src/app.js:3 | Potential command execution added | Avoid shell execution with user-controlled input... |构建完成后，可以把它配置到支持 MCP 的客户端中：
{
"mcpServers": {
"pr-guard": {
"command": "node",
"args": ["D:/github-xiang-mu/pr-guard-mcp/dist/server.js"]
}
}
}合并 PR 前做轻量安全检查
让 AI Agent 生成安全 review 评论
快速检查新增代码里是否出现明显密钥和危险 API
作为后续 GitHub PR 自动审计工具的基础版本
当前版本支持 GitHub PR URL，但还不会自动把评论写回 PR
检测规则以启发式正则为主，不能替代完整 SAST
依赖风险目前只识别新增依赖，还没有接入 npm、PyPI、Maven 漏洞库
接入 GitHub Advisory、npm audit、PyPI advisory 和 OSV
支持自动向 PR 提交 review comment
增加更多语言的危险 API 规则
| Name | Required | Description | Default |
|---|---|---|---|
| diff | Yes | Unified diff text from a pull request or git diff |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Tool has no description.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Tool has no description.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Tool has no description.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Tool has no description.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Tool has no description.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Tool has no description.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.1.0analyze_pr_diffScored across 1 tool
With only one tool, there is no possibility of overlap or confusion between tools. The lone tool is inherently distinct.
A single tool name trivially follows a consistent pattern. The verb_noun structure (analyze_pr_diff) is clear and conventional.
A single tool is too few for a server meant to guard pull requests. This feels extremely thin and likely insufficient for real-world use.
The server exposes only one operation with no description, making it impossible to assess coverage. There is no evidence of a complete lifecycle or workflow, and agents cannot perform any related actions beyond this single vague analysis.
Security reviews for coding agents: diffs checked against your org policy and live infrastructure.
Risk-scan a diff, flag AI-generated-code tells, find secrets. 5 of 7 tools need no account.
Deep security scans of repos you own from your editor: dependency CVEs, SAST, git-history secrets.
Screens public GitHub repos and PRs to generate risk maps, findings, and merge-readiness signals.
