gitlab-mr-reviewer

gitlab-mr-reviewer

AI-powered code review for GitLab merge requests that analyzes diffs, posts inline comments with severity and category, supports multi-model council review, and integrates with Kiro IDE via MCP.

Category
访问服务器

README

GitLab MR Reviewer

AI-powered code review for GitLab merge requests. Ships as an MCP (Model Context Protocol) server for use with Kiro IDE, plus a standalone CLI for CI/CD pipelines.

Features

  • GitLab MR Analysis - Fetches MR diffs, threads, and metadata via GitLab API
  • Inline Code Comments - Posts review findings directly on changed lines with severity and category
  • Multi-Model Council - Runs multiple LLMs (via AWS Bedrock) in parallel for consensus-based review
  • Thread Management - Reply, resolve, reopen discussion threads; auto-resolve fixed issues
  • Noise Filtering - Automatically skips lockfiles, generated code, build output
  • Paginated Analysis - Handles large MRs without context overflow
  • Import Resolution - Follows imports to gather context from related files
  • Repository Profiles - Learns project architecture and conventions over time
  • CI/CD Integration - Standalone CLI for automated reviews in pipelines
  • Customizable Standards - Teams can add code-review-standards/ to their repos

Quick Start

1. Install

npm install -g gitlab-mr-reviewer

2. Configure

Create a .env file or set environment variables:

GITLAB_URL=https://gitlab.com
GITLAB_PAT=glpat-xxxxxxxxxxxx

3. Use with Kiro IDE

Add to your MCP config (~/.kiro/settings/mcp.json):

{
  "mcpServers": {
    "gitlab-mr-reviewer": {
      "command": "mr-review-mcp",
      "args": [],
      "env": {
        "GITLAB_URL": "https://gitlab.com",
        "GITLAB_PAT": "glpat-xxxxxxxxxxxxxxxxxxxx"
      }
    }
  }
}

4. Use via CLI (CI/CD)

mr-review-cli --mr-url "https://gitlab.com/group/project/-/merge_requests/42"
mr-review-cli --project "group/project" --mr 42
mr-review-cli --mr-url $CI_MERGE_REQUEST_URL --dry-run
mr-review-cli --mr-url $CI_MERGE_REQUEST_URL --council

Multi-Model Council Review

Multiple AI models independently review the same MR in parallel, then findings are consolidated by consensus.

Enable Council

COUNCIL_ENABLED=true
COUNCIL_MODELS=claude-sonnet-4,claude-haiku-4.5,nova-pro
AWS_REGION=us-east-1

Environment Variables

Variable Required Default Description
GITLAB_URL Yes - GitLab instance URL
GITLAB_PAT Yes - GitLab PAT (api scope)
REVIEW_LABEL No ai-review Label added to reviewed MRs
COUNCIL_ENABLED No false Enable multi-model review
COUNCIL_MODELS No claude-sonnet-4,claude-haiku-4.5,nova-pro Models to use
AWS_REGION No us-east-1 AWS region for Bedrock
BEDROCK_MAX_TOKENS No 8192 Max tokens per model
REVIEW_MODE No deep deep (clone) or quick (diff only)

Architecture

src/
  index.ts              - Library exports
  mcp-server.ts         - MCP server class (tool implementations)
  mcp-server-stdio.ts   - STDIO transport + tool definitions
  cli-review.ts         - CLI entrypoint for CI/CD
  cache/                - MR data caching
  config/               - Retry and runtime configuration
  core/                 - Multi-model council, verification, pipeline
  modules/              - GitLab operations, analysis, comments
  types/                - TypeScript type definitions
  utils/                - Helpers (validation, errors, parsing)
power/
  POWER.md              - Kiro Power manifest
  mcp.json              - MCP server configuration
  steering/             - Review guidance files

Development

git clone https://github.com/katamarajub/gitlab-mr-reviewer.git
cd gitlab-mr-reviewer
npm install
npm run build

License

MIT

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选