经 AI Skill Hub 精选评估,Wrkr代码审计工具 获评「推荐使用」。这款MCP工具在功能完整性、社区活跃度和易用性方面表现出色,AI 评分 6.5 分,适合有一定技术背景的用户使用。
Wrkr代码审计工具 是一款遵循 MCP(Model Context Protocol)标准协议的 AI 工具扩展。通过 MCP 协议,它可以让 Claude、Cursor 等主流 AI 客户端直接访问和操作外部工具、数据源和服务,实现 AI 能力的无缝扩展。无论是文件操作、数据库查询还是 API 调用,都可以通过自然语言在 AI 对话中直接触发,极大提升生产效率。
Wrkr代码审计工具 是一款遵循 MCP(Model Context Protocol)标准协议的 AI 工具扩展。通过 MCP 协议,它可以让 Claude、Cursor 等主流 AI 客户端直接访问和操作外部工具、数据源和服务,实现 AI 能力的无缝扩展。无论是文件操作、数据库查询还是 API 调用,都可以通过自然语言在 AI 对话中直接触发,极大提升生产效率。
# 方式一:通过 Claude Code CLI 一键安装
claude skill install https://github.com/Clyra-AI/wrkr
# 方式二:手动配置 claude_desktop_config.json
{
"mcpServers": {
"wrkr------": {
"command": "npx",
"args": ["-y", "wrkr"]
}
}
}
# 配置文件位置
# macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
# Windows: %APPDATA%/Claude/claude_desktop_config.json
# 安装后在 Claude 对话中直接使用 # 示例: 用户: 请帮我用 Wrkr代码审计工具 执行以下任务... Claude: [自动调用 Wrkr代码审计工具 MCP 工具处理请求] # 查看可用工具列表 # 在 Claude 中输入:"列出所有可用的 MCP 工具"
// claude_desktop_config.json 配置示例
{
"mcpServers": {
"wrkr______": {
"command": "npx",
"args": ["-y", "wrkr"],
"env": {
// "API_KEY": "your-api-key-here"
}
}
}
}
// 保存后重启 Claude Desktop 生效
Find the bounded AI-connected software-delivery paths in your repos and org, rank the riskiest ones first, and emit offline-verifiable proof before they become unreviewed access.
Wrkr gives security and platform teams a deterministic, evidence-ready view of static AI tooling posture and gives developers a local-machine hygiene path when they want to inspect their own setup first. It discovers supported AI dev tools, MCP servers, and agent frameworks, shows what can write, highlights what to review or control first, emits proof artifacts for audits and CI, and can now render a buyer-ready static action registry summary for design-partner conversations. Wrkr stays in the static posture boundary: it does not claim runtime observation or control-layer enforcement.
Security/platform-led. Developer hygiene included. Deterministic by default.
Docs: clyra-ai.github.io/wrkr | Command reference: docs/commands/ | Examples: docs/examples/
WRKR_VERSION="v1.5.0"
go install github.com/Clyra-AI/wrkr/cmd/wrkr@"${WRKR_VERSION}"
go install github.com/Clyra-AI/wrkr/cmd/wrkr@latest
wrkr version --json
Canonical pinned install and release-parity guidance lives in docs/install/minimal-dependencies.md.
wrkr scan --my-setup --json
wrkr mcp-list --state ./.wrkr/last-scan.json --json
cp ./.wrkr/last-scan.json ./.wrkr/inventory-baseline.json
wrkr inventory --diff --baseline ./.wrkr/inventory-baseline.json --state ./.wrkr/last-scan.json --json
Use the curated scenario bundle when you want a clean evaluator-safe pass through discovery, evidence, verify, and regress without the repo-root fixture noise that shows up if you scan the Wrkr repository root directly. This bundle is intentionally risky by design. A low posture score or low first-run framework coverage is expected here and demonstrates that Wrkr is surfacing real control and evidence gaps, not that the product is failing.
wrkr scan --path ./scenarios/wrkr/scan-mixed-org/repos --json
wrkr evidence --frameworks eu-ai-act,soc2,pci-dss --state ./.wrkr/last-scan.json --output ./.tmp/wrkr-scenario-evidence --json
wrkr verify --chain --state ./.wrkr/last-scan.json --json
wrkr regress init --baseline ./.wrkr/last-scan.json --output ./.tmp/wrkr-regress-baseline.json --json
wrkr regress run --baseline ./.tmp/wrkr-regress-baseline.json --state ./.wrkr/last-scan.json --json
This curated path is the recommended fallback and demo workflow when hosted prerequisites are not ready yet or when you want to show the shipped wedge without repo-root fixture noise from Wrkr's own scenario, docs, and test fixtures. --path is explicit now: point it at a repo root when the selected directory itself carries a strong repo-root signal such as .git; point it at ./scenarios/wrkr/scan-mixed-org/repos, another immediate-child bundle root, or a bounded nested owner/repo clone when you want Wrkr to scan a deterministic repo-set. Low or zero first-run framework_coverage in this evaluator path is still an evidence-gap signal. It means the current state lacks documented controls or approvals, not that the framework mapping is unsupported.
If hosted prerequisites are still not ready yet after the evaluator-safe scenario, start with one of these deterministic local fallback paths:
wrkr scan --path ./your-repo --json
wrkr scan --my-setup --json
Use ./your-repo when the selected directory itself is the repo root. Use a bundle root like ./scenarios/wrkr/scan-mixed-org/repos when the selected directory is intentionally a set of immediate child repos.
轻量级MCP工具,专注npm安全审计。设计理念清晰,但社区认可度低,适合小规模集成试用。
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
✅ Apache 2.0 — 宽松开源协议,可商用,需保留版权声明和 NOTICE 文件,含专利授权条款。
AI Skill Hub 点评:Wrkr代码审计工具 的核心功能完整,质量良好。对于Claude Desktop / Claude Code 用户来说,这是一个值得纳入个人工具库的选择。建议先在非生产环境试用,再逐步推广。
| 原始名称 | wrkr |
| 原始描述 | 开源MCP工具:npm audit for AI agents. Wrkr scans your GitHub org and local machine for coding。⭐6 · Go |
| Topics | npm审计安全扫描AI代理代码检测Go开发 |
| GitHub | https://github.com/Clyra-AI/wrkr |
| License | Apache-2.0 |
| 语言 | Go |
收录时间:2026-05-24 · 更新时间:2026-05-30 · License:Apache-2.0 · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端