Dashboard
Your command center for managing AI agents. Chat, configure, monitor, and debug — all from one place.
Overview#
The Lobstack dashboard is where you interact with your agent. It's organized into tabs for different aspects of agent management.
Chat
The primary interface. Send messages, view AI responses with tool execution, manage multiple chat sessions, and attach files or images.
Overview
Agent status, server details, message usage meter, token usage charts, and quick actions.
Agent Config
Edit your agent's personality files (SOUL.md, PERSONA.md, RULES.md, TOOLS.md, SKILLS.md, MEMORY.md) with a live markdown editor and preview mode.
Skills
Enable and configure integrations like Discord, GitHub, Gmail, and more. Each skill adds real tools to your agent.
Sandbox
Real-time workspace file browser, workflow builder, and execution history. Watch your agent work in real-time.
Memory
View, create, and manage persistent memories. Visualized as a D3.js force-directed graph organized by category.
Logs
Real-time agent logs with color-coded severity levels. Monitor INFO, WARN, ERROR, and DEBUG messages from your agent.
Settings
Change AI model, messenger channel, server tier, region, API key, and manage your agent lifecycle.
Chat Interface#
The chat panel supports markdown rendering, code blocks with syntax highlighting, and real-time streaming responses. Your agent can execute tools during the conversation — you'll see terminal commands, file operations, and API calls happen in real time.
| Feature | Description |
|---|---|
| Sessions | Create multiple chat sessions with categories (general, task, research, support, automation, creative, debug), statuses, and pinning for quick access. |
| Session Management | Right-click any session to rename, pin/unpin, set category, mark complete, or delete. Sessions are grouped by Pinned → Active → Completed → Archived. |
| Slash Commands | Type / to see 22+ commands. Use /automate, /schedule, /trigger, /remind to create workflows from chat. Use /pin, /category, /done, /export for session management. |
| Chat-to-Automation | Create automations with natural language — e.g., /schedule every Monday at 9am check inventory creates a cron workflow automatically. |
| Image Upload | Attach images to your messages for visual analysis (model-dependent). |
| Streaming | Responses stream in real-time via WebSocket connection. |
| Tool Execution | Watch your agent run commands, call APIs, and interact with external services. |
Overview Panel#
The overview panel shows at-a-glance metrics for your agent:
- Status — Running, provisioning, or error state with uptime counter
- Message Usage — Current period usage vs. plan limit with progress bar
- Server Details — IP address, tier, region, AI model
- Token Usage — 7-day chart showing prompt and completion token consumption with per-model breakdown
- Token Conservation — Smart model routing scores message complexity (0-100) and routes to the most cost-effective model tier, saving 40-60% on API costs
- Efficiency Score — Composite metric based on error rate, completion ratio, and cost efficiency
- All Time Stats — Total messages, tokens used, active skills, and memory count
Agent Config#
The Agent Config panel lets you edit 6 markdown configuration files that define your agent's personality and behavior. Changes are pushed to your running agent immediately — no rebuild required.
| File | Purpose |
|---|---|
| SOUL.md | Core identity, purpose, and mission |
| PERSONA.md | Voice, tone, and communication style |
| RULES.md | Behavioral constraints and guardrails |
| TOOLS.md | Tool usage patterns and preferences |
| SKILLS.md | Skill-specific instructions |
| MEMORY.md | Memory management guidelines |
The editor includes a live markdown preview mode, the ability to reset any file to its default template, and unsaved-changes detection.
Configuration details
Sandbox#
The Sandbox gives you a live view into your agent's workspace. It has three sub-tabs: Files (workspace browser), Workflows (automation builder), and Executions (run history).
Logs#
The Logs panel shows real-time logs from your agent with color-coded severity levels for easy scanning.
| Level | Color | Description |
|---|---|---|
| INFO | Blue | Normal operational messages |
| WARN | Yellow | Warnings that may need attention |
| ERROR | Red | Errors that affected agent behavior |
| DEBUG | Gray | Detailed debug information |
Logs are displayed in reverse chronological order. New entries appear at the top as they come in.
Settings Panel#
From Settings, you can change:
- AI Model — Switch between 24 models from 10 providers, grouped by provider with comparison badges. Triggers a server rebuild (2-4 min).
- API Key — Provide your own API key for any provider (Anthropic, OpenAI, Google, DeepSeek, Qwen, Mistral, Groq, xAI, Moonshot). Stored on your VM, not in our database. Ollama models are self-hosted and need no key.
- Messenger — Change which channel your agent listens on (Web, Telegram, Discord, Slack, WhatsApp).
- Server Tier & Region — Upgrade or change your server specs. Creates a new VM with the new configuration.
- Rebuild — Rebuild your VM with a fresh OS install. Free, takes 2-4 minutes.
- Destroy — Delete your server. Your subscription stays active for redeployment.
Settings changes trigger rebuilds