Docs/Applications

Web Dashboard

Edit on GitHub

Accessing the Dashboard

http://localhost:18800/dashboard

The dashboard is served by the Auxiora gateway. Run auxiora start to launch the gateway and open the dashboard automatically.

Authentication

The dashboard is protected by password authentication:

  • Password hashing -- Argon2id with secure defaults (memory-hard, timing-attack resistant).
  • Sessions -- JWT tokens with configurable expiry. Stored as HTTP-only cookies.
  • First-time setup -- The setup wizard prompts you to create a password during the Vault step.

If you lose your password, use auxiora auth from the CLI to reset it.

Setup Wizard

On first launch, a 12-step wizard walks you through the full configuration:

StepNameWhat It Configures
1WelcomeIntroduction and overview
2VaultCreate or unlock the encrypted vault
3PasswordSet the dashboard access password
4IdentityYour name, role, and basic context
5PreferencesCommunication style and interaction preferences
6PersonalityChoose SOUL.md classic or The Architect engine
7AppearanceTheme, accent color, font size
8ProviderConnect your primary AI provider and model
9ChannelsEnable messaging platforms (Discord, Telegram, Slack, etc.)
10ConnectionsLink service connectors (GitHub, Notion, Home Assistant, etc.)
11ReviewSummary of all selections
12CompleteConfirmation and link to the chat interface

You can revisit any step later through the Settings pages.

Pages

Chat

The primary interaction surface. Features include:

  • Streaming responses -- Tokens arrive in real time as the model generates them.
  • Transparency footer -- Every response includes a footer showing which personality traits are active, the confidence level, and the source context used. This footer is collapsible.
  • "Why?" button -- Click to expand a full provenance breakdown: which memories were retrieved, what domain was detected, which traits were mixed, and why each decision was made.
  • Session history -- Browse and resume previous conversations from the sidebar.
  • File and image support -- Drag and drop files into the chat for analysis.

Behaviors

Create and manage automated behaviors from a visual interface:

  • Create -- Select a behavior type (scheduled, monitor, one-shot, event), configure the trigger, write the action prompt, and choose a delivery channel.
  • Edit -- Modify any behavior's configuration inline.
  • Pause / Resume -- Temporarily stop a behavior without deleting it.
  • Delete -- Permanently remove a behavior.
  • Run Now -- Manually trigger a behavior for testing.
  • Status indicators -- Each behavior shows its type, status, last run time, run count, and failure count.

See the Behaviors documentation for details on behavior types and the durable job queue.

Memory Manager

Browse and manage everything Auxiora remembers about you:

  • Browse -- View all memories in a paginated list, grouped by category (preference, fact, context, relationship, pattern, personality).
  • Search -- Full-text search across all memories.
  • Edit -- Click any memory to modify its content or category.
  • Delete -- Remove individual memories.
  • Forget -- Selectively forget an entire topic (e.g., "Forget everything about my old job"). This searches for and removes all matching memories and marks related decisions as abandoned.
  • Export -- Download all personalization data as JSON for backup or GDPR compliance.

User Profile ("About Me")

A read-only view of the user model that Auxiora has built over time. Generated by the UserModelSynthesizer, this page shows:

  • Domain expertise -- Detected areas of knowledge and skill levels based on conversation history.
  • Communication style -- How you prefer to receive information (detailed vs. brief, formal vs. casual, etc.).
  • Satisfaction trends -- Feedback patterns over time, showing what works well and what needs adjustment.
  • Active decisions -- Cross-session decisions being tracked, with follow-up dates and tags.
  • Correction patterns -- What Auxiora has learned from your corrections and how it has adapted.

This page is opt-in and explicit -- the user model is not included in prompts unless you access it through The Architect's getUserModel() method.

Settings

Ten settings pages for fine-grained control:

PageWhat It Controls
PersonalitySOUL.md tone sliders, interaction mode, personality file editing
ProviderAI provider selection, model choice, API key management
ChannelsEnable/disable messaging platforms, configure bot tokens
ConnectionsService connector setup (GitHub, Notion, Hue, etc.)
ArchitectThe Architect engine: presets, custom weights, trait configuration
AmbientAmbient intelligence: enable/disable, briefing schedule, notification preferences
AppearanceTheme (light/dark/system), accent color, font size, layout
NotificationsDesktop and in-app notification preferences, quiet hours
SecurityPassword change, trust levels, sealed auto-unseal toggle, session management
AuditView the tamper-evident audit log with chained SHA-256 hashes

Use Cases

1. First-Time Setup

Walk through the 12-step wizard to go from a fresh install to a fully configured assistant in about 10 minutes. The wizard handles vault creation, provider configuration, personality selection, and channel setup in a guided flow. No CLI knowledge required.

2. Daily Driver

Use the chat interface as your primary interaction point. The streaming responses and transparency footer give you immediate feedback on how the assistant is thinking. Pin the dashboard tab in your browser and use it alongside your other tools.

3. Memory Audit

Periodically review what Auxiora knows about you through the Memory Manager. Search for specific topics, correct any inaccuracies, and forget outdated information. Export your full personalization data at any time for backup or to transfer to another instance.

Related Documentation

  • Personality System -- The Architect engine and SOUL.md tone controls
  • Behaviors -- Behavior types, CLI commands, and durable job queue
  • Memory -- Memory categories, provenance tracking, selective forgetting
  • Vault & Security -- Authentication, trust levels, audit logging
  • CLI Reference -- Command-line alternative for all dashboard operations