AI Skill Hub 强烈推荐:Vegvisir 开源工具 是一款优质的MCP工具。AI 综合评分 8.0 分,在同类工具中表现稳健。如果你正在寻找可靠的MCP工具解决方案,这是一个值得深入了解的选择。
Vegvisir 开源工具 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。
Vegvisir 开源工具 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/Honorbound-Innovation/Vegvisir-harness
# 方式二:手动配置 claude_desktop_config.json
{
"mcpServers": {
"vegvisir-----": {
"command": "npx",
"args": ["-y", "vegvisir-harness"]
}
}
}
# 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 Vegvisir 开源工具 执行以下任务... Claude: [自动调用 Vegvisir 开源工具 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": {
"vegvisir_____": {
"command": "npx",
"args": ["-y", "vegvisir-harness"],
"env": {
// "API_KEY": "your-api-key-here"
}
}
}
}
// 保存后重启 Claude Desktop 生效
Vegvisir is a local-first agentic software development harness for people who want an AI engineering assistant that can actually work inside a repository without being handed every secret, every permission, and every memory by default.
It is not just a chat window. Vegvisir connects a model provider to an active workspace, scoped tools, durable memory, governed skills, subagents, browser evidence, approvals, verification, and a transcript that records what happened. The point is practical software work: inspect the repo, make the change, run the check, show the diff, preserve the evidence, and keep the operator in control.
Vegvisir is designed first for serious engineering workflows: code maintenance, debugging, documentation, migrations, automation, security-aware review, reverse-engineering support, browser-driven evidence capture, and long-running project sessions. It can automate work inside those boundaries, but it is not a generic "do anything on the internet" agent. The harness is intentionally shaped around workspaces, policy, memory scope, tool scope, secret isolation, and verification.
Prerequisites:
venv for Python-backed binary-intelligence components.Install native/system dependencies first when building from a fresh Linux system:
sudo bash scripts/install-system-deps.sh
This installs the core Rust/Node/Python build tools plus Tauri/WebKit/GTK desktop dependencies such as glib-2.0, Playwright/Solarium runtime libraries, and Java/Ghidra support packages where available. On Debian-like systems, ./install.sh --install-system-deps delegates to the same script.
Install the full system:
./install.sh
Install with a user HBSE broker service:
./install.sh --hbse-service user --enable-hbse-service --start-hbse-service
Install into a specific prefix:
./install.sh --prefix "$HOME/.local"
Prepare an optional low-privilege runtime account and workspace root for hardened headless deployments:
sudo ./install.sh --install-vegvisir-user --workspace-root /srv/vegvisir-workspaces
Upgrade an existing local install:
./upgrade.sh
Uninstall:
./uninstall.sh
The installer places these commands under $prefix/bin where applicable. Optional component flags such as --no-solarium, --no-biw, --no-ghidra, --no-ghidra-headless-mcp, --no-desktop, --no-skiller, --no-usrl, --no-hbse, and --no-cms-cli can omit individual systems. When Ghidra is enabled, the installer discovers an existing Ghidra installation from GHIDRA_HOME, GHIDRA_HEADLESS, or PATH and creates ghidra/analyzeHeadless wrappers that point to that installation. The upgrade script reruns install.sh from the upgraded source, and the uninstall script removes these installed commands and component trees unless data is explicitly kept.
vegvisirvegvisir-rustcms-v2hbsehbse-brokerskillerusrlsolariumbiwghidraanalyzeHeadlessghidra-headlessghidra-headless-mcpvegvisir-desktopBuild Rust crates:
cargo build --workspace
Check Rust crates:
cargo check --workspace
Run Rust tests:
cargo test --workspace -- --test-threads=1
Build and test USRL:
cd components/usrl
npm install
npm run build
npm test
Build and test Solarium:
cd components/solarium
npm install
npm run build
npm test
Vegvisir is built around a terminal workbench that keeps the conversation, tool log, session state, context budget, skills, and command surfaces visible while work is happening.

Verification output, Solarium notes, tool activity, session state, context usage, and the active input surface in one workspace-bound TUI.

Command palette, persistent agents, approvals, context tools, Skiller routing, and session tool logs during an active project session.

Long-running agent work with memory, tools, Ghidra/Skiller/CMS references, shell/test evidence, and transcript continuity.
高质量的Rust开源工具
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。
✅ MIT 协议 — 最宽松的开源协议之一,可自由商用、修改、分发,仅需保留版权声明。
总体来看,Vegvisir 开源工具 是一款质量优秀的MCP工具,在同类工具中具备一定竞争力。AI Skill Hub 将持续追踪其更新动态,建议收藏备用,结合自身场景选择合适时机引入使用。
| 原始名称 | Vegvisir-harness |
| Topics | mcprust安全 |
| GitHub | https://github.com/Honorbound-Innovation/Vegvisir-harness |
| License | MIT |
| 语言 | Rust |
收录时间:2026-06-06 · 更新时间:2026-06-06 · License:MIT · 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