{"$schema":"https://wellknown.network/schemas/agent-record-v1.json","schemaVersion":"1","id":"ag_kwkbcmtcahbq","handle":"remote-sandbox-mcp","url":"https://wellknown.network/agents/remote-sandbox-mcp","links":{"self":"https://wellknown.network/agents/remote-sandbox-mcp/record.json","html":"https://wellknown.network/agents/remote-sandbox-mcp","markdown":"https://wellknown.network/agents/remote-sandbox-mcp/record.md","api":"https://wellknown.network/api/v1/agents/remote-sandbox-mcp","status":"https://wellknown.network/api/v1/agents/remote-sandbox-mcp/status","claim":"https://wellknown.network/agents/remote-sandbox-mcp/claim","claimApi":"https://wellknown.network/api/v1/claims","claimDescriptor":"https://wellknown.network/agents/remote-sandbox-mcp/claim.json","badge":"https://wellknown.network/agents/remote-sandbox-mcp/badge.svg","openapi":"https://wellknown.network/openapi.json","history":"https://wellknown.network/api/v1/agents/remote-sandbox-mcp/history","tools":"https://wellknown.network/api/v1/agents/remote-sandbox-mcp/tools"},"ard":{"identifier":"urn:air::server:remote-sandbox-mcp","type":"application/mcp-server-card+json"},"kind":"mcp_server","declared":{"name":"remote-sandbox-mcp","summary":"MCP server for SSH-based remote sandboxes: multi-server management, persistent sessions, background tmux tasks, and GPU/CPU/memory resource monitoring","description":"# remote-sandbox-mcp\n\n[![PyPI version](https://img.shields.io/pypi/v/remote-sandbox-mcp)](https://pypi.org/project/remote-sandbox-mcp/)\n[![Python](https://img.shields.io/pypi/pyversions/remote-sandbox-mcp)](https://pypi.org/project/remote-sandbox-mcp/)\n\n一个把远程 SSH 服务器当作运行沙箱的 MCP Server，支持：\n\n- **多沙箱管理**：配置多台服务器，自动查询 CPU/内存/GPU 占用率，选出最空闲的沙箱\n- **持久 SSH 会话**：每个沙箱维持一条长连接，自动健康检查与断线重连，不再每次调用新建连接\n- **可靠的超时控制**：双层超时（远端 `timeout` 命令 + 客户端 deadline）确保超时真的生效\n- **后台长任务**：通过 `exec_bash_background` 在 tmux 中运行长任务，默认自动登记 watchdog 并返回 `watch_id`\n- **智能文件传输**：`sync_local_to_remote` 默认只内联处理小传输；超过阈值会自动切到本机后台 worker，避免单次 `tools/call` 卡住，并保留 `excludes` / `exclude_file` 规则\n- **macOS 守护监控**：本地 `launchd` watchdog 持续监控远端长任务，把状态写入 SQLite，在 SSH 连续失败后发本地通知，并在 tmux 丢失时按 checkpoint/resume 计划恢复\n- 本地文件/目录同步到远程沙箱（增量，按 `size + mtime` 跳过未变更文件）\n- 远程文件/目录同步回本地\n- 在远程沙箱执行 bash 命令并返回结果\n- 浏览远程文件列表与读取文件内容\n\n## 1. 快速添加到 Claude Code\n\n无需手动安装，直接用 `claude mcp add` 即可（依赖 [uv](https://docs.astral.sh/uv/) 自动管理运行环境）。\n\n### 单沙箱\n\n```bash\nclaude mcp add remote-sandbox \\\n  -e REMOTE_HOST=192.168.9.16 \\\n  -e REMOTE_PORT=22 \\\n  -e REMOTE_USER=your_user \\\n  -e REMOTE_PASSWORD=your_password \\\n  -- uvx remote-sandbox-mcp\n```\n\n### 多沙箱\n\n```bash\nclaude mcp add remote-sandbox \\\n  -e REMOTE_SANDBOX_LIST='[{\"name\":\"A100\",\"host\":\"192.168.9.15\",\"port\":22,\"user\":\"ubuntu\",\"password\":\"xxx\"},{\"name\":\"H100\",\"host\":\"192.168.9.16\",\"port\":22,\"user\":\"ubuntu\",\"password\":\"xxx\"}]' \\\n  -- uvx remote-sandbox-mcp\n```\n\n添加后验证：\n\n```bash\nclaude mcp list\n```\n\n> **作用域**：默认添加到当前项目（`-s project`）。如需全局可用，加 `-s user`。\n> **更新**：重新执行同一条 `claude mcp add` 命令（会覆盖旧配置），或 `uvx` 每次自动拉取最新版本。\n\n---\n\n## 2. 手动安装（Codex/Claude 通用，推荐先做）\n\n```bash\npip install remote-sandbox-mcp\n# 或\nuv tool install remote-sandbox-mcp\n```\n\n如果你使用项目虚拟环境，也可以安装到 `.venv`，然后在 Codex 配置里把 `command` 指向 `.venv/bin/remote-sandbox-mcp`。\n\n---\n\n## 3. Codex 配置指南（同一套工具）\n\n先完成第 2 节安装，再配置 Codex MCP。\n\n### 推荐配置（TOML）\n\n```toml\n[mcp_servers.\"remote-sandbox\"]\ncommand = \"/absolute/path/to/remote-sandbox-mcp\" # `which remote…","publisher":{"name":"db.xiang","url":null},"homepage":"https://github.com/DebinXiangQuantum/remote-sandbox-mcp","repository":"https://github.com/DebinXiangQuantum/remote-sandbox-mcp","version":"0.7.6","license":null,"protocols":["mcp"],"tags":["ai","claude","gpu","mcp","remote","sandbox","ssh","tmux"],"pricing":null,"endpoints":[{"url":"pypi:remote-sandbox-mcp","type":"package_pypi","auth":null,"probeable":false}],"skills":null,"tools":null,"extra":null,"attribution":{"kind":"pypi","name":"pypi","repoUrl":"pypi","summary":"pypi","version":"pypi","description":"pypi","homepageUrl":"pypi","publisherName":"pypi"}},"derived":{"capabilities":[{"slug":"code.execution","name":"Code Execution","confidence":1,"provenance":"declared"}],"categories":["code"],"language":"en"},"observed":{"status":"unknown","statusReason":"Distributed as a package to run locally; no network endpoint to check.","lastOkAt":null,"lastProbedAt":null,"statusComputedAt":null,"reliability30d":null,"latestObservations":[],"tools":null,"package":{"name":"remote-sandbox-mcp","registry":"pypi","observedAt":"2026-10-28T21:26:26.718Z","publishedAt":"2026-08-31T10:42:50.706043Z","latestVersion":"0.7.6"},"toolSurface":null,"endpointFacts":[]},"verification":{"claimed":false,"claimedAt":null,"proofs":[]},"provenance":{"sources":[{"source":"pypi","key":"remote-sandbox-mcp","url":"https://pypi.org/project/remote-sandbox-mcp/","firstSeenAt":"2026-09-10T11:26:46.013Z","fetchedAt":"2026-10-28T21:24:54.332Z","normalizedAt":"2026-10-28T21:24:54.332Z"}]},"firstSeenAt":"2026-09-10T11:26:46.013Z","updatedAt":"2026-10-28T21:26:26.718Z"}