[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-96014":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":9,"totalLinesOfCode":9,"stars":12,"forks":13,"watchers":14,"openIssues":14,"contributorsCount":9,"subscribersCount":14,"size":14,"stars1d":14,"stars7d":14,"stars30d":14,"stars90d":14,"forks30d":14,"starsTrendScore":14,"compositeScore":15,"rankGlobal":9,"rankLanguage":9,"license":9,"archived":16,"fork":16,"defaultBranch":17,"hasWiki":16,"hasPages":16,"topics":9,"createdAt":9,"pushedAt":9,"updatedAt":18,"readmeContent":19,"aiSummary":20,"trendingCount":14,"starSnapshotCount":14,"syncStatus":21,"lastSyncTime":9,"discoverSource":22},96014,"agentdock","uvwt\u002Fagentdock","uvwt","Secure MCP runtime for AI agents to operate local machines, servers, and containers with multi-device orchestration.",null,"https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock","Go",665,86,0,46.82,false,"main","2026-09-20 04:01:32","\u003Cdiv align=\"center\">\n\nEnglish | [简体中文](.\u002FREADME.zh-CN.md)\n\n\u003Cimg src=\".\u002Fdocs\u002Fassets\u002Fagentdock-logo.png\" alt=\"AgentDock logo\" width=\"128\" \u002F>\n\n# AgentDock MCP\n\n**Give AI agents secure, controlled access to every machine you operate.**\n\nOpen ChatGPT in your browser and manage multiple computers and servers from one conversation. Write code, change configuration, run commands, and deploy in the real environment where the work belongs—without consuming a dedicated Codex coding quota.\n\n\u003Ca href=\"https:\u002F\u002Ftrendshift.io\u002Frepositories\u002F136526?utm_source=trendshift-badge&utm_medium=badge&utm_campaign=badge-trendshift-136526\" target=\"_blank\" rel=\"noopener noreferrer\">\u003Cimg src=\"https:\u002F\u002Ftrendshift.io\u002Fapi\u002Fbadge\u002Ftrendshift\u002Frepositories\u002F136526\u002Fdaily?language=Go\" alt=\"uvwt%2Fagentdock | Trendshift\" width=\"250\" height=\"55\"\u002F>\u003C\u002Fa>\n\n[Documentation](https:\u002F\u002Fuvwt.github.io\u002Fagentdock-docs\u002F) · [Download](https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock\u002Freleases) · [Community](https:\u002F\u002Fqun.qq.com\u002Funiversal-share\u002Fshare?ac=1&authKey=Rp86bSzI7vqm87KoYlKawgsPZ440Ubhyezw6Qkgcn3JISwX3zXxsXkbS5598RrY5&busi_data=eyJncm91cENvZGUiOiIxMDgxMzM3MDE5IiwidG9rZW4iOiJ0Mlg1bUU1ZWtuZzF3SHJDT3pSaGsrOURIMlNYaXBlYllOUjNLZ1BUb1hzM2lJSTZjeVNldzU0ajl0SjRVZkx2IiwidWluIjoiMzIwMjA4ODAzMiJ9&data=W28mWvuqaLf_Fwnf0CgAJXuDs6l3A78V7AoWZnizPboCpKoQMzHzZ-UlluYo47U3tmIBHK2xIgWEVEJbTiGsPQ&svctype=4&tempid=h5_group_info)\n\n[![CI](https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock\u002Factions\u002Fworkflows\u002Fci.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock\u002Factions\u002Fworkflows\u002Fci.yml)\n[![GitHub Release](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fuvwt\u002Fagentdock?display_name=tag&logo=github)](https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock\u002Freleases)\n[![Docker Hub](https:\u002F\u002Fimg.shields.io\u002Fdocker\u002Fpulls\u002Fagentdockio\u002Fagentdock?logo=docker&label=Docker%20Hub)](https:\u002F\u002Fhub.docker.com\u002Fr\u002Fagentdockio\u002Fagentdock)\n[![GHCR](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGHCR-ghcr.io%2Fuvwt%2Fagentdock-2496ED?logo=docker&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock\u002Fpkgs\u002Fcontainer\u002Fagentdock)\n[![License](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Fuvwt\u002Fagentdock)](.\u002FLICENSE)\n\n\u003C\u002Fdiv>\n\n\u003Cp align=\"center\">\n  \u003Cimg\n    src=\".\u002Fdocs\u002Fassets\u002Fagentdock-multi-device.png\"\n    alt=\"AgentDock: operate multiple devices from one AI conversation\"\n    width=\"100%\"\n  \u002F>\n\u003C\u002Fp>\n\n## What is AgentDock?\n\nAgentDock is an independent tool runtime for AI agents.\n\nIt provides unified, secure, and controlled file, command, Git, Skill, MCP, browser automation, and task execution across local computers, remote servers, and containers. Connect multiple AgentDock instances to coordinate work across devices and finish multi-machine workflows in a single conversation.\n\nAgentDock does not provide a chat interface or perform model inference. It focuses on one responsibility:\n\n> Let AI agents operate real environments within explicit permission boundaries and return structured, traceable, and verifiable results.\n\n```text\n              ChatGPT \u002F Claude \u002F Codex\n                        │\n                        │ MCP (multiple instances supported)\n          ┌─────────────┼─────────────┐\n          ▼             ▼             ▼\n   ┌───────────┐ ┌───────────┐ ┌───────────┐\n   │ AgentDock │ │ AgentDock │ │ AgentDock │\n   │ Local Mac │ │ LAN Host  │ │ Cloud VPS │\n   └─────┬─────┘ └─────┬─────┘ └─────┬─────┘\n         │             │             │\n         ▼             ▼             ▼\n   Files · Shell · Git  Tunnels       Proxy · Deploy\n```\n\n## What can AgentDock do?\n\n- Manage multiple computers and servers directly from ChatGPT without repeatedly switching SSH sessions\n- Write code, modify projects, run tests, and operate Git in the real local or remote environment without depending on a dedicated coding-agent quota\n- Manage VPS hosts, Docker services, reverse proxies, and deployment configuration\n- Inspect logs, processes, ports, and actual runtime state\n- Operate authenticated web pages and macOS desktop applications\n- Connect multiple AgentDock instances and coordinate cross-device work in one conversation\n- Extend capabilities through Skills and dynamic MCP servers\n- Persist long-running task state and continue after an interruption\n- Use the same tool model across macOS, Linux, Windows, and containers\n- And more\n\n## Quick start\n\nRegular users can install AgentDock from the official package for their operating system. You do not need the source code or Go.\n\nSee [Install AgentDock](https:\u002F\u002Fuvwt.github.io\u002Fagentdock-docs\u002Fdocs\u002Fgetting-started\u002Finstall) for the complete instructions.\n\n| Platform | Documentation |\n| --- | --- |\n| Docker | [Docker installation](https:\u002F\u002Fuvwt.github.io\u002Fagentdock-docs\u002Fdocs\u002Fgetting-started\u002Fdocker) |\n| Linux | [Automated Linux installation](https:\u002F\u002Fuvwt.github.io\u002Fagentdock-docs\u002Fdocs\u002Fgetting-started\u002Flinux) |\n| Linux \u002F VPS | [Manual systemd deployment](https:\u002F\u002Fuvwt.github.io\u002Fagentdock-docs\u002Fdocs\u002Fgetting-started\u002Fvps) |\n| macOS | [macOS installation](https:\u002F\u002Fuvwt.github.io\u002Fagentdock-docs\u002Fdocs\u002Fgetting-started\u002Fmacos) |\n| Windows | [Graphical Windows installer](https:\u002F\u002Fuvwt.github.io\u002Fagentdock-docs\u002Fdocs\u002Fgetting-started\u002Fwindows) |\n\n### Choose a connection option\n\n- **Local only:** the client and AgentDock run on the same computer.\n- **Temporary public address:** ChatGPT, a phone, or another remote device needs access and no domain is ready. The address may change after the Tunnel restarts.\n- **Fixed domain:** a stable address for long-term use. Requires a Cloudflare-managed domain and Tunnel Token.\n\nAfter installation, get the MCP URL and Bearer Token or OAuth sign-in details from the control panel or terminal, then add them to the MCP, Tools, or Connectors settings in your client. Public access must keep authentication enabled. Do not include credentials in screenshots, issues, or public conversations.\n\n## Connect an AI client\n\nAgentDock exposes tools over MCP Streamable HTTP. The exact client syntax varies, but a typical configuration looks like this:\n\n```json\n{\n  \"mcpServers\": {\n    \"agentdock\": {\n      \"url\": \"http:\u002F\u002F127.0.0.1:8765\u002Fmcp\",\n      \"headers\": {\n        \"Authorization\": \"Bearer \u003CAGENTDOCK_AUTH_TOKEN>\"\n      }\n    }\n  }\n}\n```\n\n## Core capabilities\n\n### Files and commands\n\n- Read and search UTF-8 text, traverse directories, and apply structured edits\n- Atomic file writes, path boundaries, and private-directory protection\n- Command execution with timeout and output limits\n- Separate stdout, stderr, and exit status\n- Long-running command sessions, PTY, observation, input, and termination\n- Output truncation and sensitive-value redaction\n- macOS, Linux, Windows, and WSL support\n\n#### Forward selected host environment variables\n\n`exec_command` intentionally starts from a small environment instead of inheriting the complete AgentDock process environment. When a host runtime needs additional variables, configure an explicit child-to-host mapping:\n\n```bash\nexport AGENTDOCK_COMMAND_ENV_FROM_ENV_JSON='{\"NIX_LD\":\"NIX_LD\",\"NIX_LD_LIBRARY_PATH\":\"NIX_LD_LIBRARY_PATH\"}'\n```\n\nOnly mapped variables are copied. A missing host variable is skipped. Skill environment values override the mapped host value, and an explicit `exec_command.env` value overrides both.\n\nFor a systemd or OpenRC deployment, the source variables must also exist in the AgentDock service process environment. The mapping does not read a user's login shell. For example, a NixOS service using `nix-ld` should provide the current `NIX_LD` and `NIX_LD_LIBRARY_PATH` values through the service configuration together with the mapping above. Prefer declarative NixOS service configuration over snapshotting generation-specific `\u002Fnix\u002Fstore` paths into a long-lived file.\n\n### Skills and dynamic MCP\n\nOfficial and community Skill sources live in [uvwt\u002Fagentdock-skills](https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock-skills). This repository only keeps core Skills that must ship with the AgentDock runtime, including bootstrap\u002Fsecurity Skills and the built-in `agentdock-user-guide` official user guide.\n\n- Validate, install, activate, and roll back Skill packages\n- Stable, development, canary, and pinned release channels\n- Isolated environment variables and runtimes for each Skill\n- Register, enable, disable, refresh, and remove dynamic MCP servers\n- Streamable HTTP and stdio transports\n- Search tools, inspect schemas, and perform controlled calls\n- Configuration isolation between MCP servers\n\n### Native ACP\n\nAgentDock can optionally act as a native ACP client and host a local coding-agent adapter.\n\n- Desktop control panels provide presets for Codex, Claude, and Grok; host configuration controls whether ACP is enabled and which adapter is selected.\n- Use `acp_session` to create and manage sessions, `acp_prompt` to run and observe prompts, and `acp_interaction` to answer agent permission requests.\n- Optional ACP operations are available only when the connected adapter advertises the corresponding capability.\n- ACP working directories follow the host process or container security boundary rather than an AgentDock filesystem allowlist.\n\n### Browser and desktop automation\n\n- Start, close, and clean up browser sessions\n- Navigate, click, type, select, and wait\n- Inspect page text, interactive elements, errors, and network responses\n- Persist login state, use dedicated browser profiles, and capture screenshots\n- Use system Chrome and macOS desktop automation\n\n### Recoverable tasks\n\n- Persist task state\n- Define explicit goals, steps, and completion conditions\n- Record staged checkpoints\n- Track blockers and resume after interruption\n- Perform final review and evidence-based completion checks\n- Reuse workflow templates\n\n### Recall and NexusDock integration\n\nAgentDock can optionally pair with NexusDock as a multi-device aggregation entrypoint:\n\n- Long-term project memory\n- Runbooks and experience records\n- Workflow templates\n- Private notes\n- Multi-device state coordination\n- Temporary signed Artifact downloads proxied by NexusDock while the source node is online\n\n## Runtime directories\n\n| Path | Purpose |\n| --- | --- |\n| `~\u002FAgentDock` | Default working directory for relative file operations |\n| `~\u002F.agentdock` | AgentDock state, configuration, sessions, and extension data |\n\n## Ports\n\nDefault MCP URL for Docker, native installs, and local development:\n\n`http:\u002F\u002F127.0.0.1:8765\u002Fmcp`\n\nPorts are configurable. Clients must use the address defined by the actual deployment.\n\nFor public deployments, enable Bearer Token or OAuth authentication and use HTTPS. Never expose an unauthenticated MCP service to the public internet.\n\n## Development and contribution\n\nRun the full check before submitting code:\n\n```bash\nmake check\n```\n\nGitHub Actions continuously run tests, static checks, builds, and release validation.\n\nUser documentation is maintained separately in [`uvwt\u002Fagentdock-docs`](https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock-docs). Changes to user-visible behavior, configuration, installation, or tool schemas should update the matching documentation in the same change set.\n\nSubmit bugs and feature requests through [GitHub Issues](https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock\u002Fissues).\n\n## Support the project\n\n\u003Cp>If \u003Cb>AgentDock\u003C\u002Fb> helps you, please consider giving it a \u003Cb>Star\u003C\u002Fb> ⭐. Thank you for your support!\u003C\u002Fp>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth align=\"center\">WeChat\u003C\u002Fth>\n\u003Cth align=\"center\">Alipay\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd align=\"center\">\u003Cimg src=\".\u002Fdocs\u002Fassets\u002Fdonation\u002Fwechat-cropped.JPG\" alt=\"WeChat donation QR code\" height=\"200\">\u003C\u002Ftd>\n\u003Ctd align=\"center\">\u003Cimg src=\".\u002Fdocs\u002Fassets\u002Fdonation\u002Falipay.JPG\" alt=\"Alipay donation QR code\" height=\"200\">\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\n\u003C\u002Ftable>\n\n## Star History\n\n[![Star History Chart](https:\u002F\u002Fapi.star-history.com\u002Fsvg?repos=uvwt\u002Fagentdock&type=Date)](https:\u002F\u002Fstar-history.com\u002F#uvwt\u002Fagentdock&Date)\n\n## Related links\n\n- [Documentation](https:\u002F\u002Fuvwt.github.io\u002Fagentdock-docs\u002F)\n- [Documentation source](https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock-docs)\n- [GitHub Releases](https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock\u002Freleases)\n- [GitHub Container Registry](https:\u002F\u002Fgithub.com\u002Fuvwt\u002Fagentdock\u002Fpkgs\u002Fcontainer\u002Fagentdock)\n- [Docker Hub](https:\u002F\u002Fhub.docker.com\u002Fr\u002Fagentdockio\u002Fagentdock)\n- [Linux Do](https:\u002F\u002Flinux.do\u002F)\n\n## License\n\nApache License 2.0. See [LICENSE](.\u002FLICENSE).\n\n## Community\n\n[Join the QQ group (1081337019)](https:\u002F\u002Fqun.qq.com\u002Funiversal-share\u002Fshare?ac=1&authKey=Rp86bSzI7vqm87KoYlKawgsPZ440Ubhyezw6Qkgcn3JISwX3zXxsXkbS5598RrY5&busi_data=eyJncm91cENvZGUiOiIxMDgxMzM3MDE5IiwidG9rZW4iOiJ0Mlg1bUU1ZWtuZzF3SHJDT3pSaGsrOURIMlNYaXBlYllOUjNLZ1BUb1hzM2lJSTZjeVNldzU0ajl0SjRVZkx2IiwidWluIjoiMzIwMjA4ODAzMiJ9&data=W28mWvuqaLf_Fwnf0CgAJXuDs6l3A78V7AoWZnizPboCpKoQMzHzZ-UlluYo47U3tmIBHK2xIgWEVEJbTiGsPQ&svctype=4&tempid=h5_group_info)\n","AgentDock 是一个面向 AI 代理的安全 MCP（Model Context Protocol）运行时，用于在本地设备、远程服务器及容器中执行受控操作。它提供统一的文件访问、命令执行、Git 操作、技能调用、浏览器自动化和任务编排能力，并支持多实例协同以实现跨设备工作流。基于最小权限原则与显式授权机制，所有操作均结构化、可追溯、可验证。适用于 DevOps 自动化、AI 辅助运维、多机协同调试及本地化 AI 工具链集成等需要安全连接真实计算环境的场景。",2,"trending"]