{"slug": "chatpanel-a-privacy-first-ai-agent-side-panel", "title": "ChatPanel: A Privacy-First AI Agent Side Panel", "summary": "ChatPanel, an open-source browser side panel, lets users orchestrate multiple LLM agents locally with privacy features like local PII redaction and voice-to-text processing. The tool supports model-agnostic switching between Codex, Claude Code, or local models, and includes a built-in MCP hub for extensibility.", "body_md": "# ChatPanel: A Privacy-First AI Agent Side Panel\n\nPrivacy-centric AI tools usually sacrifice functionality for security, but ChatPanel manages to bridge the gap by acting as a local orchestrator for various LLM agents. Instead of being locked into a single provider, this browser side panel lets you plug in your own agents—whether that's\n\nFor anyone building a custom AI workflow, the open-source nature of the project is the biggest win. You can actually verify the privacy claims in the code rather than trusting a marketing page.\n\n4. Add your preferred MCP servers to the hub to extend the agent's capabilities.\n\n[Claude](/en/tags/claude/)Code, local models, or remote APIs—while keeping the data flow under your control.The core value here is the flexibility of the \"bridge.\" You install the extension and a few npm packages to connect your local AI environment to the browser.\n\n## Key Technical Capabilities\n\n**Model Agnostic:** You can toggle between Codex, Claude Code, or local in-browser models on a per-message basis.**Local Processing:** Voice-to-text and PII (Personally Identifiable Information) redaction happen locally, meaning sensitive data is scrubbed before it ever hits a remote server.**Context Integration:** It functions as a local note-taking app and meeting assistant, allowing you to search across your own history and notes.**Extensibility:** It includes a built-in MCP (Model Context Protocol) hub and supports CLI agents like OpenCode.**Browser Interaction:** The \"act mode\" allows the agent to interact with tools like Excalidraw directly in the browser.\n\nFor anyone building a custom AI workflow, the open-source nature of the project is the biggest win. You can actually verify the privacy claims in the code rather than trusting a marketing page.\n\n## Setup Workflow\n\nTo get this running, you generally follow these steps:\n\n1. Install the browser extension.\n\n2. Install the required bridge packages via npm:\n\n```\nnpm install -g chatpanel-bridge\n```\n\n3. Configure your API keys or point the extension to your local LLM endpoint (e.g., Ollama).4. Add your preferred MCP servers to the hub to extend the agent's capabilities.\n\nThis is a solid choice for developers who want a \"command center\" for their AI agents without shipping every single keystroke to a third-party cloud.\n\n[Next Mapbox GL JS: Turning Cities into 3D Cutscenes →](/en/threads/2448/)\n\n## All Replies （3）\n\nJ\n\nD\n\nDoes this support custom system prompts for the different agents, or is that handled externally?\n\n0\n\nA\n\nBeen using a similar setup for work docs; makes a huge difference for peace of mind.\n\n0", "url": "https://wpnews.pro/news/chatpanel-a-privacy-first-ai-agent-side-panel", "canonical_source": "https://promptcube3.com/en/threads/2464/", "published_at": "2026-07-23 18:00:44+00:00", "updated_at": "2026-07-24 02:05:40.727015+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "large-language-models", "developer-tools"], "entities": ["ChatPanel", "Claude", "Codex", "Ollama", "OpenCode", "Excalidraw"], "alternates": {"html": "https://wpnews.pro/news/chatpanel-a-privacy-first-ai-agent-side-panel", "markdown": "https://wpnews.pro/news/chatpanel-a-privacy-first-ai-agent-side-panel.md", "text": "https://wpnews.pro/news/chatpanel-a-privacy-first-ai-agent-side-panel.txt", "jsonld": "https://wpnews.pro/news/chatpanel-a-privacy-first-ai-agent-side-panel.jsonld"}}