jev-agent-quality
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Jev Agent Quality
给任何代码 Agent 使用的 Jev 质量检查工具。项目只有两部分:一个标准 MCP 服务和一份可复制的 Skill。它不绑定 Codex、Claude Code、CodeBuddy 或 DeepSeek Harness。
三步使用
1. 下载
git clone --depth 1 https://github.com/diegozhou114-cloud/jev-agent-quality.git
cd jev-agent-quality仓库已经提交可直接运行的 dist/server.js,无需安装 Node 依赖。开发或需要重新构建时使用 npm ci && npm run build。
2. 配置 key
在启动 Agent 的环境中设置:
export JEV_API_KEY="your-typesafe-key"不要把 key 写进配置文件、仓库或规则文本。
3. 配置 MCP 和 Skill
把下面的 MCP 配置复制到 Agent 的 MCP 设置中,把路径改成实际目录:
{
"mcpServers": {
"jev-agent-quality": {
"type": "stdio",
"command": "node",
"args": ["/absolute/path/to/jev-agent-quality/dist/server.js"],
"env": {"JEV_API_KEY": "${JEV_API_KEY}"}
}
}
}完整模板见 mcp.json。将 skills/jev-review/SKILL.md 复制到 Agent 的 Skill 目录,或把它的内容加入 Agent 的全局规则。这样用户不需要每次输入“使用 Jev”。
不同 Agent 的配置文件位置不同,但 MCP 都只需要这三个值:command=node、args=dist/server.js、环境变量 JEV_API_KEY。如果 Agent 支持 MCP,就可以直接使用;如果只支持规则文件,就复制 Skill 并让它调用自己的 MCP 客户端。
Related MCP server: mcp-multi-agent-code-review
它什么时候调用
Skill 会自动判断风险:跨文件、公共 API、状态/控制流、错误处理、安全、并发、测试和重构默认检查;格式、文档、拼写、依赖安装和一行机械修复默认跳过。检查只发送聚焦 diff 和必要上下文,不会自动读取或上传整个仓库。
Jev 返回结构化分数、置信度和弱项提示。Agent 负责解释、修改和测试;Jev 不改代码,也不替代测试、静态分析或人工审查。
MCP 工具
服务只提供一个只读工具:jev_review。
{
"task": "需求和验收条件",
"diff": "当前聚焦 diff",
"files": [{"path": "src/example.ts", "content": "必要的当前内容"}],
"repositoryContext": "必要的架构、约定和测试结果",
"previousEvaluation": {}
}本地开发
npm ci
npm run validate源码在 src/,构建产物在 dist/server.js。MCP 使用 stdio,适用于支持 MCP 的桌面 Agent、CLI Agent 和自建 Harness。
隐私和成本
只有明确传入 jev_review 的 task、diff、文件和上下文会发送到 TypeSafe Jev API。不要发送密钥、.env、证书、生成文件、vendor 目录或无关专有代码。按风险触发可以减少延迟和 API 成本。
License
MIT,见 LICENSE。
This server cannot be deployed
Maintenance
Related MCP Connectors
Security reviews for coding agents: diffs checked against your org policy and live infrastructure.
Read-only AI coding tools for change verification, release readiness, capacity, and guidance.
Risk-scan a diff, flag AI-generated-code tells, find secrets. 5 of 7 tools need no account.
Scores your code across 12 areas, with findings, fix prompts, and re-assessment in your AI agent
Related MCP Servers
- AlicenseNot gradedqualityAmaintenanceAI code reviews and git activity digests with machine-readable risk scoring, available as an MCP server for use within an agent session.1MIT
- FlicenseNot gradedqualityCmaintenanceRuns a multi-agent code review pipeline via MCP, analyzing git diffs for security, quality, and logic issues and returning a pass/fail verdict.-
- AlicenseNot gradedqualityBmaintenanceEnables MCP clients to run security and code review on pull requests and diffs, exposing review_pr and review_diff capabilities with local-first analyzers and LLM explanations.2MIT
- FlicenseAqualityCmaintenanceExposes Semgrep as deterministic, rule-matching analysis tools over MCP so a code-review agent can scan only the lines changed in a diff, scan whole files on demand, and check analyzer availability, version, ruleset, and limits. Results are returned as compact per-finding lines with source snippets suitable as comment anchors, along with counts of suppressed out-of-diff matches.3-