DevOps & Cloud Code Auditor MCPサーバー
コミット 2026年9月20日Provides AI assistants with static analysis tools to detect cloud cost leaks, database migration hazards, and API security vulnerabilities.
別名 Ansarii/devops-code-auditor, devops-code-auditor
- スター
- 0
- ツール
- 9
- ダウンロード数(30日間)
- —
- 最新バージョン
- 1.0.0
- 最終コミット
- 2026年9月20日
- ライセンス
- MIT
DevOps & Cloud Code Auditor MCPとは
DevOps & Cloud Code AuditorはMCPサーバーです。 Provides AI assistants with static analysis tools to detect cloud cost leaks, database migration hazards, and API security vulnerabilities. 9個のツールを提供しており、repository_url, sub_directory, statusなどがあります。
プロジェクトのREADMEから生成された概要
DevOps & Cloud Code Auditor MCP server について
Preventing production outages and unexpected cloud expenses is what DevOps & Cloud Code Auditor enables for engineers using AI assistants like Claude Desktop, Cursor, or Windsurf. By integrating this MCP server via stdio or Docker, developers can grant their AI agents the ability to perform static analysis on infrastructure and application code. The server focuses on three critical areas: cloud cost optimization for Terraform and AWS, PostgreSQL migration safety to prevent table locks, and Next.js API security to stop credential leaks or SQL injection.
Users can audit entire public Git repositories using the audit_devops_repository tool, scan local filesystem directories with audit_local_devops_directory, or pass raw code snippets directly to audit_devops_code for immediate feedback. The system provides structured JSON findings and Markdown reports that include specific fix code for issues like missing S3 lifecycle transitions, unindexed foreign keys, or unvalidated fetch calls. Because it relies entirely on AST pattern matching and regex, the tool ensures security by never executing untrusted shell commands or database connections.
できること
- 1Identify AWS cost leaksAsk your AI to scan Terraform files for expensive configurations like legacy EBS volumes or unneeded public IPs.
- 2Prevent database migration locksHave the assistant check a PostgreSQL migration script to ensure it won't halt table writes in production.
- 3Audit Next.js API securityVerify that your Next.js Server Actions include proper authentication and do not leak secrets to the client.
- 4Scan local infrastructure directoriesRun a security and cost audit on a local folder before submitting a pull request.
プロジェクトの README から作成されています。すべてのユースケースは、その引用と照らし合わせて確認されています。
DevOps & Cloud Code Auditor MCPのインストール方法
フォーマット検証日 2026年11月10日Claude CodeにDevOps & Cloud Code Auditor MCPをインストール
$ claude mcp add --transport http devops-code-auditor https://neon_innovation_lab--devops-code-auditor.apify.actor/mcpCursorにDevOps & Cloud Code Auditor MCPをインストール
~/.cursor/mcp.json
{
"mcpServers": {
"devops-code-auditor": {
"url": "https://neon_innovation_lab--devops-code-auditor.apify.actor/mcp"
}
}
}VS CodeにDevOps & Cloud Code Auditor MCPをインストール
.vscode/mcp.json
{
"servers": {
"devops-code-auditor": {
"type": "http",
"url": "https://neon_innovation_lab--devops-code-auditor.apify.actor/mcp"
}
}
}CodexにDevOps & Cloud Code Auditor MCPをインストール
~/.codex/config.toml
[mcp_servers.devops-code-auditor] url = "https://neon_innovation_lab--devops-code-auditor.apify.actor/mcp"
Gemini CLIにDevOps & Cloud Code Auditor MCPをインストール
~/.gemini/settings.json
{
"mcpServers": {
"devops-code-auditor": {
"httpUrl": "https://neon_innovation_lab--devops-code-auditor.apify.actor/mcp"
}
}
}DevOps & Cloud Code Auditor MCPツール 9
READMEから検出repository_urlsub_directorystatustotal_findingsfindingsmarkdown_reportcodefile_typedirectory_path
| repository_url | — | |
| sub_directory | — | |
| status | — | |
| total_findings | — | |
| findings | — | |
| markdown_report | — | |
| code | — | |
| file_type | — | |
| directory_path | — |
README
GitHubで見るDevOps & Cloud Code Auditor (MCP Server)
A production-ready Model Context Protocol (MCP) Server providing AI coding assistants (Claude Desktop, Cursor, Windsurf, Cline) with automated static analysis tools to prevent production outages, cloud cost leaks, and security vulnerabilities.
🌟 Capabilities & Audited Rule Sets
This server inspects codebases across three mission-critical engineering vectors:
-
Cloud Cost Leaks (Terraform / HCL / AWS):
EBS_GP2: Identifies legacygp2EBS volumes (gp3provides equal durability with ~20% baseline cost reduction and higher baseline IOPS).PUBLIC_IP: Flags unneededassociate_public_ip_address = trueincurring hourly AWS IPv4 charges ($0.005/hr/IP).CW_NO_RETENTION: Catchesaws_cloudwatch_log_groupmissing retention policies (preventing infinite unbudgeted log ingestion costs).S3_NO_LIFECYCLE: Detects S3 buckets lacking Glacier lifecycle transitions.RDS_MULTIAZ_NONPROD&RDS_PROVISIONED_IOPS_NONPROD: Flags doubled database infrastructure costs on development and staging environments.
-
PostgreSQL Migration Table Lock Hazards:
INDEX_NOT_CONCURRENT: DetectsCREATE INDEXmissingCONCURRENTLY(which acquires anEXCLUSIVElock and halts table writes in production).NOT_NULL_VOLATILE_DEFAULT: CatchesADD COLUMN ... NOT NULL DEFAULT <func()>rewriting every table row under anACCESS EXCLUSIVElock.FK_NO_INDEX: Flags unindexed foreign key constraints causing cascading table-level sequential lock scans on parent updates/deletes.FOR_UPDATE_UNBOUNDED: CatchesSELECT ... FOR UPDATEmissingLIMITorSKIP LOCKED, preventing deadlocks.
-
OWASP & Full-Stack Next.js API Security:
NEXT_PUBLIC_SECRET_LEAK: Flags sensitive tokens and private API keys prefixed withNEXT_PUBLIC_that leak into public client browser bundles.NEXTJS_SERVER_ACTION_NO_AUTH: Flags Next.js 14/15 Server Actions exporting mutations without explicit session auth verification.SQLI_RAW_INTERPOLATION: Identifies raw SQL query templates using string interpolation instead of parameterized inputs.SSRF_UNVALIDATED_FETCH: Catches external fetch calls consuming user inputs without protocol/hostname validation.
🛠️ MCP Tools Exposed
This server implements the official Model Context Protocol specification (tools/list and tools/call):
1. audit_devops_repository
Clones a public Git repository shallowly without executing hooks, performs AST and regex static syntax audits, and returns structured findings and a formatted Markdown report.
- Parameters:
repository_url(string, required): Public Git URL of the repository to audit (e.g.https://github.com/example/cloud-infra).sub_directory(string, optional): Specific subdirectory within the repository to inspect (e.g.terraform/orprisma/migrations/).
- Returns:
status: "success" | "error"total_findings: Number of detected issuesfindings: Array of detailed vulnerability records (file, line number, rule ID, severity, message, fix code)markdown_report: Human-readable summary table
2. audit_devops_code
Directly audits a raw code snippet of Terraform (HCL), PostgreSQL migration (SQL), or Next.js API/Server Action code.
- Parameters:
code(string, required): The source code string to audit.file_type(string, optional, default: "auto"): One of"terraform","postgres","nextjs", or"auto".
- Returns: Structured JSON findings with exact drop-in code fixes.
3. audit_local_devops_directory
Audits a local filesystem directory on the host machine before submitting a Pull Request or deploying to staging.
- Parameters:
directory_path(string, required): Absolute filesystem path to the target folder.
- Returns: Structured JSON findings and severity breakdown.
🚀 Client Configuration & Quickstart
1. Claude Desktop Setup
Add the server to your claude_desktop_config.json:
{
"mcpServers": {
"devops-code-auditor": {
"command": "python3",
"args": ["/path/to/devops-code-auditor/server.py"]
}
}
}
2. Cursor IDE / Windsurf Setup
In Cursor, navigate to Settings → Features → MCP → Add New MCP Server:
- Name:
devops-code-auditor - Type:
command - Command:
python3 /absolute/path/to/devops-code-auditor/server.py
3. Docker Execution
Run the isolated container via Docker:
{
"mcpServers": {
"devops-code-auditor": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"ghcr.io/ansarii/devops-code-auditor"
]
}
}
}
💻 Example Agent Usage
Once connected, your AI assistant can evaluate migration files and cloud manifests on demand:
User: "Check this PostgreSQL migration before I run it on production:
CREATE INDEX idx_orders_user_id ON orders (user_id);"
Claude (Tool Call):
audit_devops_code({
"code": "CREATE INDEX idx_orders_user_id ON orders (user_id);",
"file_type": "postgres"
})
Result:
{
"status": "success",
"total_findings": 1,
"findings": [
{
"suite": "postgres-locks",
"rule": "INDEX_NOT_CONCURRENT",
"severity": "high",
"file": "migration.sql",
"line": 1,
"message": "CREATE INDEX on \"orders\" without CONCURRENTLY blocks table writes during build.",
"fix": "CREATE INDEX CONCURRENTLY idx_orders_user_id ON orders (user_id);"
}
]
}
🔒 Security Guarantee: Zero Execution
This server executes 100% static analysis using AST pattern matching and regex verification. It never executes untrusted shell commands, arbitrary Python code, or database connections on your machine.
📜 License
MIT License — Copyright (c) 2026 Neon Innovation Lab. Maintained by Neon Innovation Lab.
DevOps & Cloud Code Auditor MCPのよくある質問
DevOps & Cloud Code Auditor MCPをClaude Codeに追加するには
ターミナルでclaude mcp add --transport http devops-code-auditor https://neon_innovation_lab--devops-code-auditor.apify.actor/mcpを実行します。すべてのプロジェクトで利用するには--scope userを追加します。
DevOps & Cloud Code Auditor MCPをCursorに追加するには
~/.cursor/mcp.jsonを開き、上記のインストールセクションに示すCursor設定を使ってmcpServersの下にサーバーを追加します。
DevOps & Cloud Code Auditor MCPが提供するツールは
9個のツールを提供しており、repository_url, sub_directory, status, total_findings, findingsなどがあります。
What kind of code can this server audit?
It can audit Terraform (HCL), PostgreSQL migrations, and Next.js API or Server Action code.
Does the server execute my code?
No, the server uses 100% static analysis via AST pattern matching and regex, ensuring no untrusted commands or database connections are run.
Can I audit a public GitHub repository?
Yes, the audit_devops_repository tool allows you to clone and inspect public Git repositories shallowly.