Soul-driven AI agent with permission-hardened tools, token budgets, and multi-channel access. Runs 24/7 from CLI, Telegram or More.
---
title: "Web Dashboard"
sidebar_label: "Web Dashboard"
---
# Web Dashboard
Mercury Web is a React single-page application that provides a visual interface for managing every part of the agent — chat, memory, tasks, code, and configuration. It runs locally at `http://127.0.0.1:6174`.
## Enabling the Dashboard
Enable the web dashboard using either method:
- Run `mercury doctor` and follow the prompts
- Set `web.enabled: true` in your `mercury.yaml` configuration
## Configuration
| Option | Default | Description |
|--------|---------|-------------|
| `web.enabled` | `false` | Enable or disable the web dashboard |
| `web.port` | `6174` | Port the dashboard listens on |
| `MERCURY_PORT` | `6174` | Environment variable override for port |
The server binds to `127.0.0.1` only and is not exposed to the network. The web server lifecycle is tied to the Mercury process — it stops when Mercury stops.
## Authentication
The dashboard is protected by username/password authentication.
There is no default password. `mercury setup` asks you for one (at least 10 characters) or generates a strong one and shows it once. The username is `mercury`. Lost it? Run `mercury web-reset-password`.
You will be redirected to `/login` for any unauthenticated request. After logging in, an HTTP-only session cookie is set. Sign out from **Settings → Account** or by closing the session in your browser.
You can change your credentials from **Settings → Account**.
## Install as a PWA
Mercury Web is a Progressive Web App. In Chrome, Edge, or Safari, look for the **Install** icon in the address bar (or **File → Install Mercury Agent**) to add it to your launcher. The PWA runs in its own window without browser chrome.
## Pages
### Status
Landing page at `/`. Shows Mercury's current process state, active provider, daily token usage, recent activity, and quick links to other sections.
### Chat (`/chat` and `/chat/:threadId`)
A multi-thread chat interface with SSE streaming.
- **Thread sidebar** — Lists all conversations. Click any thread to switch; the URL updates to `/chat/<id>` so you can bookmark or share links.
- **New Chat** button — Creates a new thread and navigates to its URL.
- **Rename** — Use the pencil beside a conversation to choose a name (up to 120 characters). Save applies it; Cancel or Escape keeps the existing name. Your name survives restarts and automatic title generation.
- **Delete** — Hover a thread (or open it) to reveal a trash icon. Confirmation dialog prevents accidents.
- **Model switcher** — Top of the chat header. Pick any configured provider; the choice is per-thread.
- **Code Mode toggle** — Switches the agent into coding mode for the active thread (file edits, shell calls allowed under your permission policy).
- **Streaming** — Assistant tokens arrive via Server-Sent Events; you can interrupt mid-stream.
- **Attachments** — Use the paperclip, paste files, or drop them into the composer. Up to four files can accompany a message, including a message with no caption. Images and PDFs keep their existing model requirements; UTF-8 `.txt`, `.md`, and `.markdown` files work with every model.
Text attachments are limited to 256 KiB each. Mercury includes the first 16,000 characters per file, up to 32,000 characters across the turn, with a filename and an explicit truncation notice. The complete original remains available from the message's file link. Binary files, invalid UTF-8, and control characters other than tabs and line breaks are rejected. Markdown is passed as text and downloaded as plain text; embedded HTML is not rendered.
The same text-file support is available in Mercury Desktop and through the chat and bot message APIs using `attachments: [{ name: "notes.md", data: "<base64 UTF-8>" }]`. Channel adapters that already supply document paths use the same bounded text reader.
### Tasks (`/tasks`)
View and manage background and scheduled tasks the agent has created or that you queued via Telegram or chat.
### Board (`/board` and `/board/:boardId`)
The Kanban Boards view for visually planning and executing agent work. See the [Kanban Boards](./kanban-boards) page for full details.
### Workspace (`/workspace`)
A browser-based IDE for editing project files, running terminal commands, and managing Git — all without leaving Mercury. See the [Coding Workspace](./coding-workspace) page for full details.
### Second Brain
Mercury's memory and knowledge views — see the [Second Brain](../reference/second-brain) reference.
- **Memory** (`/second-brain/memory`) — Browse, search, edit, promote, or prune memory items. Filter by scope (`conscious`, `subconscious`) and by tag.
- **Persons** (`/second-brain/persons`, `/second-brain/persons/:id`) — People Mercury has met, with their facts, last-seen timestamps, and relationships.
- **Goals** (`/second-brain/goals`) — Long-running goals with progress, sub-tasks, and linked memories.
- **Graph** (`/second-brain/graph`) — Force-directed graph visualization of memories, persons, and goals with their relationships.
### Configure
- **Providers** (`/providers`) — List configured LLM providers, their priority order, and switch the default.
- **Permissions** (`/permissions`) — Review and adjust the YOLO/strict permission policy for tools.
- **Skills** (`/skills`) — Install, list, and remove agent skills. See the [Skills](../reference/skills) page.
- **Schedules** (`/schedules`) — Manage recurring or one-shot scheduled commands.
- **Usage** (`/usage`) — Per-day token usage charts.
- **Profit Sharing** (`/profit-sharing`) — Cooperative revenue-share settings (if enabled).
### Settings (`/settings`)
Three sections:
- **Defaults** — Default provider and daily token budget.
- **Account** — Change your username or password.
- **Appearance** — Light / Dark / System theme toggle.
The agent personality is no longer editable from the web UI — modify it in `mercury.yaml` if you need to customise it.
## Tech Stack
- **Server:** Hono on Node.js
- **Frontend:** React 19 + Vite + Tailwind CSS, shadcn/ui components
- **State:** Zustand for client state, SSE for streaming
- **PWA:** vite-plugin-pwa