$git clone https://github.com/agentify-sh/desktopAgentify Desktop is a local control center for AI web sessions. It lets MCP-capable tools such as Codex, Claude Code, and OpenCode use the AI subscriptions you are already signed into, while keeping browser state, files, and automation on your machine.
| 1 | # Agentify Desktop |
| 2 | |
| 3 | Agentify Desktop is a local control center for AI web sessions. It lets MCP-capable tools such as Codex, Claude Code, and OpenCode use the AI subscriptions you are already signed into, while keeping browser state, files, and automation on your machine. |
| 4 | |
| 5 | ## What It Does |
| 6 | |
| 7 | - Opens a local Agentify Control Center. |
| 8 | - Manages signed-in browser sessions for ChatGPT, Claude, Perplexity, Gemini, Google AI Studio, and Grok. |
| 9 | - Exposes MCP tools for querying a tab, reading a page, navigating, uploading files, saving artifacts, and reusing stable tab keys. |
| 10 | - Supports parallel tabs so different agents or tasks can use separate sessions. |
| 11 | - Packs local repo/file context into prompts when requested. |
| 12 | - Saves generated images/files locally so they can be reused in follow-up prompts. |
| 13 | |
| 14 | ## Example Prompts After MCP Setup |
| 15 | |
| 16 | Once Agentify Desktop is running and registered with your MCP client, you can ask for workflows like: |
| 17 | |
| 18 | - “Use Agentify with key `repo-triage` to ask ChatGPT for a second opinion on this bug, then compare its answer with your own analysis.” |
| 19 | - “Open a Perplexity tab with key `research-auth-flow` and research current OAuth best practices for desktop apps.” |
| 20 | - “Send this implementation plan to Claude in a separate Agentify tab and summarize any risks it finds.” |
| 21 | - “Use Agentify to generate three UI concept images, save the images as artifacts, and return the local file paths.” |
| 22 | - “Open Grok and ChatGPT in separate Agentify tabs, ask both to review this API design, then compare the tradeoffs.” |
| 23 | - “Pack this repo into context, ask ChatGPT to identify risky files, and save the conversation under a stable tab key for follow-ups.” |
| 24 | - “Read the current ChatGPT page through Agentify and turn the conversation into actionable TODOs.” |
| 25 | |
| 26 | ## Requirements |
| 27 | |
| 28 | - Node.js 20 or newer |
| 29 | - An MCP-capable CLI if you want tool integration: Codex, Claude Code, or OpenCode |
| 30 | |
| 31 | ## Supported Sites |
| 32 | |
| 33 | - `chatgpt.com` |
| 34 | - `claude.ai` |
| 35 | - `perplexity.ai` |
| 36 | - `aistudio.google.com` |
| 37 | - `gemini.google.com` |
| 38 | - `grok.com` |
| 39 | |
| 40 | ## Preferred Install And Run |
| 41 | |
| 42 | Start the desktop GUI without cloning this repo: |
| 43 | |
| 44 | ```bash |
| 45 | npx @agentify/desktop |
| 46 | ``` |
| 47 | |
| 48 | Equivalent explicit GUI command: |
| 49 | |
| 50 | ```bash |
| 51 | npx @agentify/desktop gui |
| 52 | ``` |
| 53 | |
| 54 | If you prefer a global install: |
| 55 | |
| 56 | ```bash |
| 57 | npm install -g @agentify/desktop |
| 58 | agentify-desktop |
| 59 | ``` |
| 60 | |
| 61 | If you want the older repo-clone and local source workflow, use [DEVELOPMENT_FROM_SOURCE.md](DEVELOPMENT_FROM_SOURCE.md). |
| 62 | |
| 63 | ## MCP Server |
| 64 | |
| 65 | Run the MCP server over stdio: |
| 66 | |
| 67 | ```bash |
| 68 | npx @agentify/desktop mcp |
| 69 | ``` |
| 70 | |
| 71 | Show newly-created browser tabs while debugging: |
| 72 | |
| 73 | ```bash |
| 74 | npx @agentify/desktop mcp --show-tabs |
| 75 | ``` |
| 76 | |
| 77 | With a global install: |
| 78 | |
| 79 | ```bash |
| 80 | agentify-desktop-mcp |
| 81 | agentify-desktop-mcp --show-tabs |
| 82 | ``` |
| 83 | |
| 84 | ## Register With MCP Clients |
| 85 | |
| 86 | Codex: |
| 87 | |
| 88 | ```bash |
| 89 | codex mcp add agentify-desktop -- npx -y @agentify/desktop mcp |
| 90 | ``` |
| 91 | |
| 92 | Claude Code: |
| 93 | |
| 94 | ```bash |
| 95 | claude mcp add --transport stdio agentify-desktop -- npx -y @agentify/desktop mcp |
| 96 | ``` |
| 97 | |
| 98 | OpenCode config example: |
| 99 | |
| 100 | ```json |
| 101 | { |
| 102 | "mcp": { |
| 103 | "agentify-desktop": { |
| 104 | "type": "local", |
| 105 | "command": ["npx", "-y", "@agentify/desktop", "mcp"], |
| 106 | "enabled": true |
| 107 | } |
| 108 | } |
| 109 | } |
| 110 | ``` |
| 111 | |
| 112 | Use `--show-tabs` at the end of the command while debugging: |
| 113 | |
| 114 | ```bash |
| 115 | codex mcp add agentify-desktop -- npx -y @agentify/desktop mcp --show-tabs |
| 116 | ``` |
| 117 | |
| 118 | ## First Run |
| 119 | |
| 120 | 1. Start the app: |
| 121 | |
| 122 | ```bash |
| 123 | npx @agentify/desktop |
| 124 | ``` |
| 125 | |
| 126 | 2. In the Control Center, create or show a ChatGPT tab. |
| 127 | |
| 128 | 3. Sign in to the target vendor in the browser window. |
| 129 | |
| 130 | 4. Register the MCP server with your CLI. |
| 131 | |
| 132 | 5. Ask your MCP client to use Agentify: |
| 133 | |
| 134 | ```text |
| 135 | Use Agentify Desktop with tab key repo-triage. |
| 136 | Ask ChatGPT to summarize this repo in 8 bullets and list the top 3 risky areas to change first. |
| 137 | Return the answer and keep the tab key stable for follow-ups. |
| 138 | ``` |
| 139 | |
| 140 | The core loop is: |
| 141 | |
| 142 | - keep a real signed-in browser session open locally |
| 143 | - call it from an MCP client |
| 144 | - reuse a stable tab key across follow-up prompts |
| 145 | |
| 146 | ## Useful MCP Tools |
| 147 | |
| 148 | The MCP server registers `agentify_*` tools, including: |
| 149 | |
| 150 | - `agentify_query`: |